Build critique request and the story behind it.
This year I purchased an APC UPS because of my flakey connection to the grid. Ominous read clicks coming from my boot drive cemented that decision. I am happy to say that not only did I avoid my biannual unexpected shutdowns, the UPS also sensed transient RF spikes. It begins to click, switching to the battery and back to mains, even though there is no visible disruption to service, although the emergency lighting in the hallway was triggered and beeped itself on. I decided to purchase a case and build a new desktop. I waited till now because I was loosely following Sarah Sharp, of Intel and listened to her comments on USB. Version 3.0 is RMA as of Jan this year and her general comment is that hardware development has peaked and so the immediate future is in developing software. Here is the case I chose. I take delivery tomorrow. http://www.coolermaster.com/case/lan-box-haf-series/haf-xb-evo/ Newegg has this combo on offer. Intel Core i5-7500 Kaby Lake 4-Core 3.4 GHz CPU w water cooling unit ASUS STRIX H270F GAMING MB G.SKILL 2 x 8G DDR4 As a creative anachronist or purposefull recycler, I haven't built a complete system from all new parts in a dozen years. I thought I'd get my hand in before arthritis takes over my hardware chops completely. I'd appreciate any warnings and caveats regarding this combo. I'm still window shopping but what I'm looking for is stability and longevity over gaming. This combo has DDR4 and USB 3.1. I'm kind of fixed on the usb but I'm not married to the ram. So suggestions which lower the cost are most appreciated. I have a 650w Corsair p/s which is ok but a big factor for me is the ambient noise and recommendations on a quiet supply would be appreciated. Noise is why I'm trying out water cooling. From the reviews on the case I'll probably have to replace the case fans, but who knows I might get lucky. I follow with interest the posts on UEFI, secure boot, SElinux, USB, udisks, udev, mobile linux and systemd. Now is my time to see if I have learned anything. Thanks to all posters. Regards -- Russell Sent by K-9 Mail
| From: Russell via talk <talk@gtalug.org> | Intel Core i5-7500 Kaby Lake 4-Core 3.4 GHz CPU w water cooling unit I get the feeling (i.e. have not checked) that Intel has introduced the 8th generation at similar price points to the 7th, but the 8th has two more cores at most levels. So I'd look at i5-8xxx. The current motherboards for 8th gen are the expensive ones and cheaper ones will be introduced (price based on chipset). Intel probably made the 8th gen a better value than the 7th was at its day is that AMD became credible again. If you are going to use a separate video card, consider looking at AMD processors. If you are just going to use integrated graphics, you have to go Intel at this time. | I follow with interest the posts on UEFI, secure boot, SElinux, USB, | udisks, udev, mobile linux and systemd. Now is my time to see if I have | learned anything. It mostly just works. Unless it doesn't. Or you try to do something different.
On November 15, 2017 4:54:50 PM EST, "D. Hugh Redelmeier via talk" <talk@gtalug.org> wrote:
| From: Russell via talk <talk@gtalug.org>
| Intel Core i5-7500 Kaby Lake 4-Core 3.4 GHz CPU w water cooling unit
I get the feeling (i.e. have not checked) that Intel has introduced the 8th generation at similar price points to the 7th, but the 8th has two more cores at most levels. So I'd look at i5-8xxx.
A quick check gets me this combo. https://www.asus.com/ca-en/Motherboards/PRIME-Z370-P/ and https://ark.intel.com/products/126685/Intel-Core-i5-8600K-Processor-9M-Cache... If I keep the same water cooling and take a slight hit on the ram clockwise, this costs out at $27 more than my original. Not too bad for the two extra cores. This bios has all the overclocking geegaws and gimcracks any tweaker would love to have in fron of them. I think I basically gave up optical audio out and the disco flashing led features, I hope.
The current motherboards for 8th gen are the expensive ones and cheaper ones will be introduced (price based on chipset).
Intel probably made the 8th gen a better value than the 7th was at its day is that AMD became credible again.
If you are going to use a separate video card, consider looking at AMD processors. If you are just going to use integrated graphics, you have to go Intel at this time.
Integrated for now. I'm not dealing with any intensive video but I appreciate quality audio and noise cancellation. This board seem to have good internal conditioning. I'm going to check some audio reviews.
| I follow with interest the posts on UEFI, secure boot, SElinux, USB, | udisks, udev, mobile linux and systemd. Now is my time to see if I have | learned anything.
It mostly just works. Unless it doesn't. Or you try to do something different.
I think the biggest improvement, for an end user like myself is; the entry level knowledge needs are much lower and the growing user base promotes quicker hardware driver development. Also there is now a much wider base to draw your information from, than even five years ago. The downside is that for most desktop uses, once I install any GNU Linux for someone, I rarely get service calls. I pulled in my travelling tech for home users shingle a long time ago. Not enough trade to support it. Also I got pretty tired of saying I dont do windows. Thanks for the great tips.
--- Talk Mailing List talk@gtalug.org https://gtalug.org/mailman/listinfo/talk
-- Russell Sent by K-9 Mail
On Wed, Nov 15, 2017 at 07:24:41PM -0500, Russell via talk wrote:
A quick check gets me this combo.
Looks quite good, although for me personally I find the lack of displayport concerning. Displayport trivially converts to DVI or HDMI, while the other way around isn't the case. Displayport support daisy chaining and high resolution, while DVI is essentially useless for modern displays. Just something that might matter. The HDMI port is also 1.4b, not 2.0, so it is limited to 30Hz at 4k resolution. The -A version of the board adds displayport and has a max of 4096x2304@60Hz and keeps the other two ports. Also has intel rather than realtek for the network port. I guess that probably means it costs more too. Yeah $219 vs $175. Well that's annoying for the budget.
and
https://ark.intel.com/products/126685/Intel-Core-i5-8600K-Processor-9M-Cache...
If I keep the same water cooling and take a slight hit on the ram clockwise, this costs out at $27 more than my original. Not too bad for the two extra cores. This bios has all the overclocking geegaws and gimcracks any tweaker would love to have in fron of them. I think I basically gave up optical audio out and the disco flashing led features, I hope.
If quiet is the goal, make sure to check the reviews on the water cooler since I found a number were quite noisy a couple of years ago when I built a machine for a friend that had water cooling. At the time, the intel brand one was apparently quite noisy. This may well have changed by now. I remember being surprised that so many water coolers were rather noisy. -- Len Sorensen
On November 16, 2017 11:32:09 AM EST, lsorense@csclub.uwaterloo.ca wrote:
On Wed, Nov 15, 2017 at 07:24:41PM -0500, Russell via talk wrote:
A quick check gets me this combo.
Looks quite good, although for me personally I find the lack of displayport concerning. Displayport trivially converts to DVI or HDMI, while the other way around isn't the case. Displayport support daisy chaining and high resolution, while DVI is essentially useless for modern displays. Just something that might matter. The HDMI port is also 1.4b, not 2.0, so it is limited to 30Hz at 4k resolution.
The -A version of the board adds displayport and has a max of 4096x2304@60Hz and keeps the other two ports. Also has intel rather than realtek for the network port. I guess that probably means it costs more too. Yeah $219 vs $175. Well that's annoying for the budget.
Thanks, I'm sold. I hadn't fully considered forward compatibility in respect of the display. The improved features justify the new price point. Although I'm a little wary of the Realtek S1220A codec. There is kernel support for it in 4.11 and people have made it work. It does have an integrated preamp and pop filter, that works for me
and
https://ark.intel.com/products/126685/Intel-Core-i5-8600K-Processor-9M-Cache...
If I keep the same water cooling and take a slight hit on the ram
clockwise, this costs out at $27 more than my original. Not too bad for the two extra cores. This bios has all the overclocking geegaws and gimcracks any tweaker would love to have in fron of them. I think I basically gave up optical audio out and the disco flashing led features, I hope.
If quiet is the goal, make sure to check the reviews on the water cooler since I found a number were quite noisy a couple of years ago when I built a machine for a friend that had water cooling. At the time, the intel brand one was apparently quite noisy. This may well have changed by now. I remember being surprised that so many water coolers were rather noisy.
I said noise, but its a bit more than that. A home users computers biggest enemy is hosehold dust and microscopic particles of cooking oil. I hate trying to clean monster heat sinks. With radiator cooling I can do ordinary cleaning of the fan from the outside and the actual cooling unit is all smooth surfaces, much easier to clean the sticky dross. I took delivery of the case today. Just in time, I lost the boot sector of the remaing drive on my desktop. I had shutdown and pulled my active data and archive disks, no joy on reboot. No significant data gone, just have to boot from a live usb distro to see if I can recover a couple of docs. Many thanks Lennart and Hugh. I revamped my cost calculations and pushed the ram clock back up to DDR 4000, but with half the the amount. I can upgrade that later, if it is even necessary. The new config +A MB and RAM, added about $80 on my build, but the added features make it very worthwhile. The only thing left for to consider is the SSD for system files. Cheers. -- Russell Sent by K-9 Mail
I'm sorry that I didn't mention this earlier. One useful site is https://ca.pcpartpicker.com/ Mind you, I've only used it a few times. This is supposed to find the best online price for the parts you are looking for. It does not consider physical stores, but many of them have an online presence. Interesting factoid: it no longer considers NCIX and related sites due to NCIX's apparent death spiral. One interesting feature is that you can leave your parts collection on pcpartpicker and point others at that list. With convenient links, it is kind of handy for discussions. My son and I partially collaborated on a system for him using this facility (well, the US site because he is in the US). Now, more than a year later, I can still reference that list when he asks me a quetion about some part. (Linux support for his AMD video card has been a problem.)
On Fri, Nov 17, 2017 at 11:19:26AM -0500, D. Hugh Redelmeier via talk wrote:
Interesting factoid: it no longer considers NCIX and related sites due to NCIX's apparent death spiral.
NCIX is having a death spiral? I guess I haven't been buying any parts in the last year so I hadn't paid attention. That explains why I couldn't find the hours for the toronto stores on their website. Apparently they are not there anymore. Looks like 3 more BC stores are closing today in about an hour. Looks like they just have 3 stores left in BC then. That's a shame. At least Canada Computers seems to be running fine. -- Len Sorensen
On November 17, 2017 11:19:26 AM EST, "D. Hugh Redelmeier via talk" <talk@gtalug.org> wrote:
I'm sorry that I didn't mention this earlier.
One useful site is https://ca.pcpartpicker.com/ Mind you, I've only used it a few times.
This is supposed to find the best online price for the parts you are looking for. It does not consider physical stores, but many of them have an online presence.
Nice web tool. I love the voltage and instant compatibility checker. If it's right, it's what office automation is all about. If it's wrong, caveat emptor, they are a third party.
Interesting factoid: it no longer considers NCIX and related sites due to NCIX's apparent death spiral.
One interesting feature is that you can leave your parts collection on pcpartpicker and point others at that list. With convenient links, it is kind of handy for discussions.
Here's my current list. I've ordered the storage already. Both sales ended soon. I also added a modular 500w P/S and a cooling fan pack so I can try out push pull static radiant cooling. It might help to keep fan noise levels down, or not, who can say. I'm going to drop my current PSU back in a 4u rack I keep for emergencies in order to attempt to ressurect some data. https://ca.pcpartpicker.com/list/RQTkFd Adding in the cost of the LAN case I'm $250 over my price point. However that point didn't figure in the cost of driving from store to store and checking out the deals in 1994 and then ordering and driving back to pick it up. This was all time and money for me back in the day. In 1994 it cost me $50 dollars a day to start my truck, not counting gas.
My son and I partially collaborated on a system for him using this facility (well, the US site because he is in the US). Now, more than a year later, I can still reference that list when he asks me a quetion about some part. (Linux
One thing that confuses me is manufacturing reporting conventions. For instance the DDR4-4000 modules are described as a Column Access Strobe latency of 19. This CPU reports support for DDR4-2666 CAS 15. This number refers to the onboard cache, I think? I went with a single 8 gig DDR4-2666 module, just to be safe. I wish I was better at math. I attribute some of my confusion to having been frightened off math by my having viewed the machine math room at my vocational school. All those heavy welded tables sorting punch cards, all that noise concentrated in a small room brrrr, gave me the willies. I took electrical wiring (knob and tube) and mechanical drafting instead. I can calculate angles and areas using the tables of a rafter square and I'm getting better at troubleshooting DSDT acpi issues. However, I imagine those issues will be as obsolete as knob and tube wiring soon, if not already. support for his AMD video card has
been a problem.) --- Talk Mailing List talk@gtalug.org https://gtalug.org/mailman/listinfo/talk
Offhand I'm planning to put /home and /opt on the 3TB drive but if you have any recommendations for an optimal partitioning scheme for the SSD, in order to reduce writes as an effort to prolong service life, I'd appreciate it. Thanks, -- Russell
| From: Russell via talk <talk@gtalug.org> | Adding in the cost of the LAN case I'm $250 over my price point. What's a LAN case? Local Area Network? | However | that point didn't figure in the cost of driving from store to store and | checking out the deals in 1994 and then ordering and driving back to | pick it up. This was all time and money for me back in the day. In 1994 | it cost me $50 dollars a day to start my truck, not counting gas. It's not very meaningful to compare 1994 and 2017 prices. And the retail marketplace is way different. Compared with 2012 prices, current prices seem high to me. - Moore's law is flagging so progress is slower. This make the replacement market a lot smaller. - fewer stores trying to grow marketshare in a no-longer growing market. - Intel completely dominates the processor market and prices accordingly (with a slight dip due to AMD's Ryzen) - Cryptocurrency Miners have bid up the price of AMD video cards - RAM prices are high. Supposedly due to fabs switching to producing flash memory. But I suspect lack of competition due to consolidation or collusion. - HDD prices are static. I paid less for 3T Seagates from Newegg three years ago (Black Friday 2014 price: $89.99). | One thing that confuses me is manufacturing reporting conventions. For | instance the DDR4-4000 modules are described as a Column Access Strobe | latency of 19. This CPU reports support for DDR4-2666 CAS 15. This | number refers to the onboard cache, I think? No, it is not about the on-chip CPU cache. The signals going in and out of RAM modules are complicated. You could look at a timing diagram to understand them (I have, a few times, but it doesn't stick in my head). What follows is from possibly defective memory (in my head). Latency is about delay from sending a signal until getting a result. Except that for RAM it is broken into stages. CAS tells the module that the computer has placed the column number portion of the address on the address pins (addresses are broken into half (column and row addresses), and each half is presented to the module's address pins in turn. CAS latency says how many clock cycles come between the computer asserting CAS and the result showing up. Your CPU can support as few as 15. Your RAM actually takes 19. No problem. Memory module characteristics are available in a little ROM on the module itself. The computer firmware queries all those ROMs and configures the computer's memory controller(s) to match. Sometimes firmware setup screens let you override those setting but I would not do so. | I went with a single 8 gig DDR4-2666 module, just to be safe. No need. In fact, your computer would probably be faster with two. But probably not enough to matter. Modern memory systems are quite a lot more flexible than old ones. Memories don't need to be matched (but a slow one will probably slow down a fast one). | I wish I was better at math. This doesn't seem like math to me. But then my math environment isn't normal. Apparently "math phobia" is a thing that one can absorb from the environment. There's a whole literature about this and the damage is does. | I attribute some of my confusion to having | been frightened off math by my having viewed the machine math room at my | vocational school. All those heavy welded tables sorting punch cards, | all that noise concentrated in a small room brrrr, gave me the willies. That sounds like "unit record" equipment. Not really good for math. Generally used for accounting up until the early or mid 1960s. The few classrooms that had them probably kept them a long time since they had been quite expensive. Keypunches lasted perhaps 15 years longer than the other unit record equipment. Band saws and open-hearth forge fans would be louder. (In my shop class, the forge used coal which we had to coke before using.) | Offhand I'm planning to put /home and /opt on the 3TB drive but if you | have any recommendations for an optimal partitioning scheme for the SSD, | in order to reduce writes as an effort to prolong service life, I'd | appreciate it. If you do the arithmetic (necessarily guestimates) SSDs will probably last as long as you'll want to use them, and longer. In my limited experience, SSDs typically fail more abruptly and completely than HDDs. So backups are even more important. Since I'm lazy about backups, I generally put the OS on the SSD and my stuff on the HDD. At least when I have both SSD and HDD. But fewer of my computers have HDDs at all. And I really should do more backups. I find modern desktops (Windows and Linux) cry out for SSD performance. I'm setting up CentOS 7 on a little computer that came with a 32G SSD. That's fine for the OS + 10G swap (I added a 2T HDD for /var and /home). Your SSD is 512G so it would be a waste to use it just for the OS.
On November 19, 2017 12:36:09 PM EST, "D. Hugh Redelmeier via talk" <talk@gtalug.org> wrote:
| From: Russell via talk <talk@gtalug.org>
| Adding in the cost of the LAN case I'm $250 over my price point.
What's a LAN case? Local Area Network?
| However | that point didn't figure in the cost of driving from store to store and | checking out the deals in 1994 and then ordering and driving back to | pick it up. This was all time and money for me back in the day. In 1994 | it cost me $50 dollars a day to start my truck, not counting gas.
It's not very meaningful to compare 1994 and 2017 prices. And the retail marketplace is way different.
Compared with 2012 prices, current prices seem high to me.
In 2012 the CAD/US exchange rate was 1:1; in 94 it was .73 and 2016 .75. In my mind I used a sort of social equity metric in that; what I'm buying now is positioned in the same middle ground of what is currently affordable over the counter, as it was with my last fully new parts computer in 1994. <snip Moore's law etc.>
Modern memory systems are quite a lot more flexible than old ones. Memories don't need to be matched (but a slow one will probably slow down a fast one).
| I wish I was better at math.
This doesn't seem like math to me. But then my math environment isn't normal.
Apparently "math phobia" is a thing that one can absorb from the environment. There's a whole literature about this and the damage is does.
I failed math horribly in grade 9. So did most of my class. I guess to hide the fact that maybe there was a problem with the teacher, or the methods, they graded us on the curve till enough people had passed. I still only scored 36%
| I attribute some of my confusion to having | been frightened off math by my having viewed the machine math room at my | vocational school. All those heavy welded tables sorting punch cards,
| all that noise concentrated in a small room brrrr, gave me the willies.
That sounds like "unit record" equipment. Not really good for math. Generally used for accounting up until the early or mid 1960s. The few
classrooms that had them probably kept them a long time since they had been quite expensive. Keypunches lasted perhaps 15 years longer than the other unit record equipment.
Band saws and open-hearth forge fans would be louder.
I took printing, at least the machines weren't working full bore all the time. I hand set type from the California Job Case, used the Linotype machine, negative etched emulsions to create plates for forms and made the companion plates for used for printing the table metadata on the form. Also darkroom photography and a bunch of other interesting markup tools.
(In my shop class, the forge used coal which we had to coke before using.)
Wow, you can coke coal ... cool eh?
| Offhand I'm planning to put /home and /opt on the 3TB drive but if you | have any recommendations for an optimal partitioning scheme for the SSD, | in order to reduce writes as an effort to prolong service life, I'd | appreciate it.
If you do the arithmetic (necessarily guestimates) SSDs will probably last as long as you'll want to use them, and longer.
In my limited experience, SSDs typically fail more abruptly and completely than HDDs. So backups are even more important.
Since I'm lazy about backups, I generally put the OS on the SSD and my stuff on the HDD. At least when I have both SSD and HDD. But fewer of my computers have HDDs at all. And I really should do more backups.
I find modern desktops (Windows and Linux) cry out for SSD performance.
I'm setting up CentOS 7 on a little computer that came with a 32G SSD.
That's fine for the OS + 10G swap (I
Interesting, how did you decide on 10G, this doesn't look like the standardly recommended1.5 - 2xRAM. added a 2T HDD for /var and
/home). Your SSD is 512G so it would be a waste to use it just for the OS.
I was thinking of partitioning /boot using existing Grub2 inode limits, then use 512-byte inodes on the rest of the filesystem. Not exactly sure why except for a vague idea that, having headroom in SElinux policy metadata headers, if used on a dedicated IPV6 LAN, might work out ok. I've never actually tried to tune a Linux filesystem, thought I might have at it now, see how it goes. _
--- Talk Mailing List talk@gtalug.org https://gtalug.org/mailman/listinfo/talk
-- Russell
| From: Russell via talk <talk@gtalug.org> | On November 19, 2017 12:36:09 PM EST, "D. Hugh Redelmeier via talk" <talk@gtalug.org> wrote: | >I'm setting up CentOS 7 on a little computer that came with a 32G SSD. | > | >That's fine for the OS + 10G swap (I added a 2T HDD for /var and | >/home). | | Interesting, how did you decide on 10G, this doesn't look like the | standardly recommended1.5 - 2xRAM. The recommendations are apparently all obsolete. I do think that it is good to have at least as much swap as RAM so you could have a chance at hibernation. But then again, I don't know if you actually need more to preserve already-swapped stuff. And generally I don't hibernate boxes. So it is sort of based on my superstition. My box currently has 2G of RAM but has had 10G at times. | >Your SSD is 512G so it would be a waste to use it just for the OS. | | I was thinking of partitioning /boot using existing Grub2 inode limits, | then use 512-byte inodes on the rest of the filesystem. Not exactly sure | why except for a vague idea that, having headroom in SElinux policy | metadata headers, if used on a dedicated IPV6 LAN, might work out ok. | I've never actually tried to tune a Linux filesystem, thought I might | have at it now, see how it goes. What are grub2 inode limits? Apparently grub 0.97 limited inodes to 128 bytes. A lot of tuning is aimed at HDD rather than SSD. SSDs tend to do a lot behind the back of the computer. The try to pretend to have HDD geometries but the actual logical structure of the disk is way different to the physical structure. For SSDs: - try to leave a certain amount of empty/unused space. This reduces the amount of shuffling that the drive does (autonomously and invisibly). How much free space? I don't know. But behaviour is surely non-linear (hockey-stick curve) so you want to stay left of the bend. I'd guess 5% is fine. You could take your hint from how much "over-provisioning" is done on commercial-grade SSDs vs consumer grade ones. The best way to leave space for this purpose is to leave a portion of the drive outside any partition. But the second best way is sometimes more useful: avoid filling all the filesystems. Here's a random blurb on over-provisioning <http://www.samsung.com/semiconductor/minisite/ssd/downloads/document/Samsung_SSD_845DC_04_Over-provisioning.pdf> It has a few precise numbers from tests but it does not disclose the tests. So it is half-scientific. - use fstrim(8) once in a while to inform the disk of the blocks in deleted files. How often? It depends on how much you have deleted. I typically do it once a month or once a week, but I have no support for those frequencies. Perhaps LVM prevents fstrim from working. Certainly that used to be the case.
On November 19, 2017 11:17:54 PM EST, "D. Hugh Redelmeier via talk" <talk@gtalug.org> wrote:
| From: Russell via talk <talk@gtalug.org>
| On November 19, 2017 12:36:09 PM EST, "D. Hugh Redelmeier via talk" <talk@gtalug.org> wrote:
| >I'm setting up CentOS 7 on a little computer that came with a 32G SSD. | > | >That's fine for the OS + 10G swap (I added a 2T HDD for /var and | >/home). | | Interesting, how did you decide on 10G, this doesn't look like the | standardly recommended1.5 - 2xRAM.
The recommendations are apparently all obsolete.
I do think that it is good to have at least as much swap as RAM so you could have a chance at hibernation. But then again, I don't know if you actually need more to preserve already-swapped stuff. And generally I don't hibernate boxes. So it is sort of based on my superstition.
My box currently has 2G of RAM but has had 10G at times.
| >Your SSD is 512G so it would be a waste to use it just for the OS. | | I was thinking of partitioning /boot using existing Grub2 inode limits, | then use 512-byte inodes on the rest of the filesystem. Not exactly sure | why except for a vague idea that, having headroom in SElinux policy | metadata headers, if used on a dedicated IPV6 LAN, might work out ok.
| I've never actually tried to tune a Linux filesystem, thought I might
| have at it now, see how it goes.
What are grub2 inode limits?
Apparently grub 0.97 limited inodes to 128 bytes.
Grub2 apparently raised those limits to 256 bytes.
A lot of tuning is aimed at HDD rather than SSD. SSDs tend to do a lot behind the back of the computer. The try to pretend to have HDD geometries but the actual logical structure of the disk is way different to the physical structure.
I've wondered about this, does a SSD Cascade fail or can it somehow be journaled to exclude bad cells and the boot sector and or superblock be logically reallocated to other sectors. Also from what I've read so far, it is deletes and not writes which stress a SSD the most. I've bookmarked this overview. https://www.mydigitaldiscount.com/everything-you-need-to-know-about-slc-mlc-...
For SSDs:
- try to leave a certain amount of empty/unused space. This reduces the amount of shuffling that the drive does (autonomously and invisibly). How much free space? I don't know. But behaviour is surely non-linear (hockey-stick curve) so you want to stay left of the bend. I'd guess 5% is fine. You could take your hint from how much "over-provisioning" is done on commercial-grade SSDs vs consumer grade ones.
Overproisioning that's the correct term for my intended unallocated space. Thanks, that will make reading up on things easier.
The best way to leave space for this purpose is to leave a portion of the drive outside any partition. But the second best way is sometimes more useful: avoid filling all the filesystems.
Here's a random blurb on over-provisioning <http://www.samsung.com/semiconductor/minisite/ssd/downloads/document/Samsung_SSD_845DC_04_Over-provisioning.pdf> It has a few precise numbers from tests but it does not disclose the tests. So it is half-scientific.
- use fstrim(8) once in a while to inform the disk of the blocks in deleted files. How often? It depends on how much you have deleted. I typically do it once a month or once a week, but I have no support for those frequencies.
Perhaps LVM prevents fstrim from working. Certainly that used to be the case.
I think the answer to that might be the typical yes and no, in that the feature is apparently under development. https://bugs.centos.org/view.php?id=10191
--- Talk Mailing List talk@gtalug.org https://gtalug.org/mailman/listinfo/talk
-- Russell
| From: Russell via talk <talk@gtalug.org> I seem to be good at giving too-late advice. The coming thing in SSDs is NVMe. <https://en.wikipedia.org/wiki/NVM_Express> These live directly on the PCI bus and thus get rid of the SATA bottleneck. But most current SSDs aren't significantly bottlenecked by SATA. You need OS support (in Linux for 5 years but improved more recently) and hardware support (prehaps only firmware for booting from NVMe). I bet that your motherboard, chipset, and CPU chip support NVMe. NVMe SSDs cost more than SSDs with SATA, mSATA, and m.2 connectors. I don't think that any of my machines supports NVMe.
On November 19, 2017 1:18:00 PM EST, "D. Hugh Redelmeier via talk" <talk@gtalug.org> wrote:
| From: Russell via talk <talk@gtalug.org>
I seem to be good at giving too-late advice.
Good advice never comes too-late. It is after all good advice and if not me, I'm pretty sure someone else could use it sometime in the future. I'm pretty good at talking in generalized marketing hyperbole myself. My LAN box is actually this Cooler Master case. https://m.newegg.ca/products/N82E16811119265 The 4U rack I have in storage has, at last check, a perfectly service able Asus M3A78-EM mb. I'm going to use that for any data recovery I wind up having to try to do. I have one outlet left on the UPS so I'm probably going to then run that unit headless and store HD video from my phone which I've down-sampled. There's a hawk flying around my neighbourhood. As of yet I haven't been able to catch him in flight chasing crows. So I'll have to start recording as soon as I see him, but I can't even email short videos without down-sampling them. Also this way I'll extend the use of my current IDE drives.
The coming thing in SSDs is NVMe. <https://en.wikipedia.org/wiki/NVM_Express>
These live directly on the PCI bus and thus get rid of the SATA bottleneck. But most current SSDs aren't significantly bottlenecked by SATA.
You need OS support (in Linux for 5 years but improved more recently) and hardware support (prehaps only firmware for booting from NVMe). I bet that your motherboard, chipset, and CPU chip support NVMe.
I'll check into that for future reference. It seems, given your previous comment on SSD's being prone to complete failure, I should put the kybosh on my plan to leave some unallocated space on that drive as LVM headroom for /boot.
NVMe SSDs cost more than SSDs with SATA, mSATA, and m.2 connectors.
I don't think that any of my machines supports NVMe. --- Talk Mailing List talk@gtalug.org https://gtalug.org/mailman/listinfo/talk
-- Russell
| From: Russell via talk <talk@gtalug.org> | My | LAN box is actually this Cooler Master case. | | https://m.newegg.ca/products/N82E16811119265 Ahh. "LAN" as in "LAN party". So the box is meant to look cool and be portable. | There's a hawk flying around my neighbourhood. | | As of yet I haven't been able to catch him in flight chasing crows. I seem to remember it usually going the other way. A bunch (murder?) of crows hound a hawk. | So | I'll have to start recording as soon as I see him, but I can't even | email short videos without down-sampling them. Depending on the phone, you could probably scp the files while your phone is on your LAN through WiFi. That should not have a size limit. | It seems, given your previous | comment on SSD's being prone to complete failure, I should put the | kybosh on my plan to leave some unallocated space on that drive as LVM | headroom for /boot. I'm not sure what that means. I don't have much LVM experience. Do you mean: space outside of the LVM-managed physical partition? To be used for /boot? Since you can fairly easily re-install an OS, /boot doesn't seem particularly precious.
On Sun, Nov 19, 2017 at 01:18:00PM -0500, D. Hugh Redelmeier via talk wrote:
| From: Russell via talk <talk@gtalug.org>
I seem to be good at giving too-late advice.
The coming thing in SSDs is NVMe. <https://en.wikipedia.org/wiki/NVM_Express>
These live directly on the PCI bus and thus get rid of the SATA bottleneck. But most current SSDs aren't significantly bottlenecked by SATA.
You need OS support (in Linux for 5 years but improved more recently) and hardware support (prehaps only firmware for booting from NVMe). I bet that your motherboard, chipset, and CPU chip support NVMe.
NVMe SSDs cost more than SSDs with SATA, mSATA, and m.2 connectors.
I don't think that any of my machines supports NVMe.
Some NVMe come as m.2 modules. The board being looked at certainly supports m.2 NVMe modules. For example the Samsung 960 EVO M.2. Those things are fast and go significantly faster than any SATA port could do. A 4x PCIe 3.0 connection gives a lot more bandwidth than a single SATA link, and NVMe avoids the SATA protocol overhead too. -- Len Sorensen
On Sun, Nov 19, 2017 at 09:35:45AM -0500, Russell via talk wrote:
One thing that confuses me is manufacturing reporting conventions. For instance the DDR4-4000 modules are described as a Column Access Strobe latency of 19. This CPU reports support for DDR4-2666 CAS 15. This number refers to the onboard cache, I think?
Well if running at 2666 speed, 15 clock cycles sounds close to what 19 clock cycles at 4000 speed would be (given how much shorter those clock cycles would be). 4000/19 = 210 2666/15 = 177 Not quite the same, but similar. Maybe the default values are just conservative or that 4000 ram is very very agresive in its timings.
I went with a single 8 gig DDR4-2666 module, just to be safe.
Yeah super fast ram can sometimes be tricky to work with and I am not convinced it really makes much difference. Slower ram is usually also easier to upgrade to more ram later. Often there are limits on the really fast stuff in terms of how many slots are in use and matching speeds between modules.
I wish I was better at math. I attribute some of my confusion to having been frightened off math by my having viewed the machine math room at my vocational school. All those heavy welded tables sorting punch cards, all that noise concentrated in a small room brrrr, gave me the willies.
I took electrical wiring (knob and tube) and mechanical drafting instead. I can calculate angles and areas using the tables of a rafter square and I'm getting better at troubleshooting DSDT acpi issues. However, I imagine those issues will be as obsolete as knob and tube wiring soon, if not already.
support for his AMD video card has
been a problem.)
Yes ATI/AMD has long had a history of great hardware and crappy drivers. I keep hoping someday they will figure out the driver issue and then seeing that no they still haven't.
Offhand I'm planning to put /home and /opt on the 3TB drive but if you have any recommendations for an optimal partitioning scheme for the SSD, in order to reduce writes as an effort to prolong service life, I'd appreciate it.
Well make sure to use 1MB alignment for the start of partitions. I think the partition tools do that by default these days anyhow. Wear leveling ought to mean it doesn't matter how you partition though. If you want to reduce wear, don't write to it. Could mean writing logs elsewhere. Of course you do have to write quite a lot to wear out an SSD so for most people it might not be a problem. I certainly don't give it any thought on my own machines. -- Len Sorensen
| From: Lennart Sorensen via talk <talk@gtalug.org> | Wear leveling ought to mean it doesn't matter how you partition though. | If you want to reduce wear, don't write to it. True, but it is more complicated. Underneath the facade of a normal HDD, an SSD does a bunch of tricky things. Terminology (mine): virtual block: what the disk host adapter and OS sees. Just like a block on an HDD. real block: a chunk of flash that can hold one virtual block erase block: the smallest unit of flash that can be erased. - an erase block contains a lot of real blocks. Think roughly a megabyte. The collection of real blocks within it is fixed. - only erased real blocks can be written to. And only once before they are erased again. - real blocks can be in one of three states: + free (not representing any virtual block but not erased) + erased (not representing any virtual block, erased) Note: an erased real block is not an erase block (it will be inside an erase block). + in-use (representing a virtual block) - in the real hardware, you can never update a block in place. So when a program writes to a virtual block, a real, erased block is written and some book-keeping is done. If the write was to a virtual block that was represented by a real block, that real block becomes free: there is no way for the computer to reference it, so it need not be preserved. - the SSD firmware keeps track of erased blocks. When it runs out, it does a garbage collect phase to find unused blocks. If it finds that a whole erase-block is full of free blocks, it will erase that block and add it to the free pool. But that isn't normal. Normally, an erase block is like swiss cheese and the good stuff has to be moved to an erased block to allow their former erase block to be erased. As you can see, a write to a block might precipitate as much as 1MiB of actual writes. That's called "write amplification" and it is can wear out SSDs quite seriously. And it will slow things down a lot. - how does the drive firmware learn that a physical block is free? + a block on an SSD is born free + a write to a virtual block will cause a write to a newly allocated physical block AND implicitly make the old physical block free (but not erased!) + deleting a file on an SSD causes its virtual blocks to be free, but the SSD firmware does not know that until a trim command tells it. Consequences: - having a lot of free physical blocks cuts down on write amplification - the effect is non-linear - to increase the number of free blocks + use trim * fstrim(8) * trimm option to mount + allocate less of the disk drive for OS use. But, if it isn't a new disk, you have to tell the SDD firmware that the free space is free. I don't know how to do that.
On 2017-11-20 03:44 PM, D. Hugh Redelmeier via talk wrote:
True, but it is more complicated.
Underneath the facade of a normal HDD, an SSD does a bunch of tricky things.
Terminology (mine):
virtual block: what the disk host adapter and OS sees. Just like a block on an HDD.
real block: a chunk of flash that can hold one virtual block
erase block: the smallest unit of flash that can be erased.
- an erase block contains a lot of real blocks. Think roughly a megabyte. The collection of real blocks within it is fixed.
- only erased real blocks can be written to. And only once before they are erased again.
- real blocks can be in one of three states:
+ free (not representing any virtual block but not erased)
+ erased (not representing any virtual block, erased) Note: an erased real block is not an erase block (it will be inside an erase block).
+ in-use (representing a virtual block)
- in the real hardware, you can never update a block in place. So when a program writes to a virtual block, a real, erased block is written and some book-keeping is done.
If the write was to a virtual block that was represented by a real block, that real block becomes free: there is no way for the computer to reference it, so it need not be preserved.
- the SSD firmware keeps track of erased blocks. When it runs out, it does a garbage collect phase to find unused blocks. If it finds that a whole erase-block is full of free blocks, it will erase that block and add it to the free pool.
But that isn't normal. Normally, an erase block is like swiss cheese and the good stuff has to be moved to an erased block to allow their former erase block to be erased.
As you can see, a write to a block might precipitate as much as 1MiB of actual writes. That's called "write amplification" and it is can wear out SSDs quite seriously. And it will slow things down a lot.
- how does the drive firmware learn that a physical block is free?
+ a block on an SSD is born free
+ a write to a virtual block will cause a write to a newly allocated physical block AND implicitly make the old physical block free (but not erased!)
+ deleting a file on an SSD causes its virtual blocks to be free, but the SSD firmware does not know that until a trim command tells it.
Consequences:
- having a lot of free physical blocks cuts down on write amplification
- the effect is non-linear
- to increase the number of free blocks
+ use trim * fstrim(8) * trimm option to mount
+ allocate less of the disk drive for OS use. But, if it isn't a new disk, you have to tell the SDD firmware that the free space is free. I don't know how to do that.
One of the better write ups that I've seen about SSDs in general and and over-provisioning specifically: https://www.seagate.com/ca/en/tech-insights/ssd-over-provisioning-benefits-m... Cheers, Jamon
On November 20, 2017 3:44:37 PM EST, "D. Hugh Redelmeier via talk" <talk@gtalug.org> wrote:
| From: Lennart Sorensen via talk <talk@gtalug.org>
| Wear leveling ought to mean it doesn't matter how you partition though. | If you want to reduce wear, don't write to it.
True, but it is more complicated.
Underneath the facade of a normal HDD, an SSD does a bunch of tricky things.
<trim the middle>
Consequences:
- having a lot of free physical blocks cuts down on write amplification
- the effect is non-linear
- to increase the number of free blocks
+ use trim * fstrim(8) * trimm option to mount
+ allocate less of the disk drive for OS use.
How would you suggest swap discard policy be handled? I have read about some people with fast systems setting 0 swap without issues, it's not a recommended practice tho. swapon, --discard[=policy] Enable swap discards, if the swap backing device supports the discard or trim operation. This may improve performance on some Solid State Devices, but often it does not. The option allows one to select between two available swap discard policies: --discard=once to perform a single-time discard operation for the whole swap area at swapon; or --discard=pages to asynchronously discard freed swap pages before they are available for reuse. If no policy is selected, the default behavior is to enable both discard types. The /etc/fstab mount options discard, discard=once, or discard=pages may also be used to enable discard flags. I had tentatively planned a very minimal swap on the SSD and to place any /extraswap on a HDD should it prove to be necessary. I was also going to try out using systemd units rather than fstab entries. https://www.commandlinux.com/man-page/man5/systemd.swap.5.html Black friday happened yesterday on a Monday (go figure) and my Newegg cart dropped my CPU choice as N/A, also Amazon refunded my 3TB drive choice, also probably N/A. Right now I have a delivery label for the SSD, so I'm pretty sure it's on its way. Sometimes I install three or four times until I'm satisfied with the balance of a partitioning scheme. That's usually on reclaimed or RMA new old stock. I'm not sure that I want to format and reformat a new SSD that way, so I think I'd like to try and get this one up with optimal housekeeping on the first go, if possible, so any comments are most helpful.
But, if it isn't a new disk, you have to tell the SDD firmware that the free space is free. I don't know how to do that. --- Talk Mailing List talk@gtalug.org https://gtalug.org/mailman/listinfo/talk
Thanks. -- Russell
| From: Russell via talk <talk@gtalug.org> | swapon, --discard[=policy] Enable swap discards, if the swap backing | device supports the discard or trim operation. Thanks. Interesting information (that I've trimmed). "discard" is the SCSI term for what ATA calls "trim". I have never addressed this. Apparently the default behaviour looks OK. | Black friday happened yesterday on a Monday (go figure) No, these are the pre-Black-Friday offers. BF is coming the day after US Thanksgiving (this Thursday). "Cyber Monday" is the Monday after that. I'm getting a lot of ads. | and my Newegg | cart dropped my CPU choice as N/A, also Amazon refunded my 3TB drive | choice, also probably N/A. Carts are usually time-limited. In fact, until you "buy", they can disappear. Or worse, as you've found, they can even back out of a sale that you have already concluded (but that's not as routine). Was the drive "sold by amazon" or another party? Places to look online for CPUs include Mike's, Canada Computers, Memory Express, Newegg.ca, Amazon.ca. pcpartpicker probably does that but doesn't always know about sales right away. Sometimes you need to look at the ads to find the deal. | Sometimes I install three or four times until I'm satisfied with the | balance of a partitioning scheme. That's usually on reclaimed or RMA new | old stock. I'm not sure that I want to format and reformat a new SSD | that way, so I think I'd like to try and get this one up with optimal | housekeeping on the first go, if possible, so any comments are most | helpful. blkdiscard(8) looks useful. You could apply it to the whole disk before you start (re-)partitioning. mkfs.ext4 has a "discard" option that defaults on. So maybe everything "just works".
On November 21, 2017 11:23:21 AM EST, "D. Hugh Redelmeier via talk" <talk@gtalug.org> wrote:
| From: Russell via talk <talk@gtalug.org>
| swapon, --discard[=policy] Enable swap discards, if the swap backing | device supports the discard or trim operation.
Thanks. Interesting information (that I've trimmed).
"discard" is the SCSI term for what ATA calls "trim".
I've always found it a weird polyglot convention that ATA drives are mounted as /dev/sdx. I know that has to do more with historical convergence of increased ATA transfer rates with scsi parallelism, but it's still kind of confusing.
I have never addressed this. Apparently the default behaviour looks OK.
| Black friday happened yesterday on a Monday (go figure)
No, these are the pre-Black-Friday offers. BF is coming the day after US Thanksgiving (this Thursday). "Cyber Monday" is the Monday after that.
I'm getting a lot of ads.
| and my Newegg | cart dropped my CPU choice as N/A, also Amazon refunded my 3TB drive | choice, also probably N/A.
Carts are usually time-limited. In fact, until you "buy", they can disappear. Or worse, as you've found, they can even back out of a sale that you have already concluded (but that's not as routine).
Was the drive "sold by amazon" or another party?
I can't tell from my order details. I should have read the ad more carefully. The SSD made it from NJ to TO, delivery date tentatively, is tomorrow. The MB is coming via Canada Post, no date yet. Case, MB & SSD = $625CAD
Places to look online for CPUs include Mike's, Canada Computers, Memory Express, Newegg.ca, Amazon.ca. pcpartpicker probably does that but doesn't always know about sales right away. Sometimes you need to look at the ads to find the deal.
Thanks. There is available on Newegg an i7-8700 Coffee Lake 6-Core 3.2 GHz (4.6 GHz Turbo) LGA 1151 (300 Series) 65W which comes with cooling, not water tho, for about $75 more than the i5-8600k which came without a heat sink, so I'm waffling between them a bit.
| Sometimes I install three or four times until I'm satisfied with the | balance of a partitioning scheme. That's usually on reclaimed or RMA new | old stock. I'm not sure that I want to format and reformat a new SSD | that way, so I think I'd like to try and get this one up with optimal
| housekeeping on the first go, if possible, so any comments are most | helpful.
blkdiscard(8) looks useful. You could apply it to the whole disk before you start (re-)partitioning.
Thanks, I'll have a look. I guess the drive will have some sort of MS file system to deal with firstly.
mkfs.ext4 has a "discard" option that defaults on.
So maybe everything "just works".
Hope so. I do have a tendency to overthink stuff. I'd hate to have myself as a regular customer, I'd drive myself bonkers. Thanks for all the pointers.
--- Talk Mailing List talk@gtalug.org https://gtalug.org/mailman/listinfo/talk
-- Russell
On Tue, Nov 21, 2017 at 04:41:55PM -0500, Russell via talk wrote:
I've always found it a weird polyglot convention that ATA drives are mounted as /dev/sdx. I know that has to do more with historical convergence of increased ATA transfer rates with scsi parallelism, but it's still kind of confusing.
Well given ATAPI is SCSI over ATA, and hence to support anything other than harddisks, you need to support SCSI commands, it seems like making ATA disk command just be another flavor of scsi in the kernel made sense. I am not sure how SATA fits in given SAS supports SATA but not the other way around. I believe SATA still uses ATA commands, unless it is not a harddisk in which cases it uses SCSI through ATAPI as it did on IDE. I have seen suggestions that it makes more sense to think of /dev/sdX as storage disk X rather than scsi disk X. Doesn't help for /dev/sg which is scsi generic, so we probably should just live with it and not worry about the naming as long as it works. Not like /dev/hdX ever meant IDE. I believe it was used for IDE, MFM, and a few other types that are probably no longer supported. -- Len Sorensen
On November 22, 2017 10:41:25 AM EST, lsorense@csclub.uwaterloo.ca wrote:
On Tue, Nov 21, 2017 at 04:41:55PM -0500, Russell via talk wrote:
I've always found it a weird polyglot convention that ATA drives are mounted as /dev/sdx. I know that has to do more with historical convergence of increased ATA transfer rates with scsi parallelism, but it's still kind of confusing.
Well given ATAPI is SCSI over ATA, and hence to support anything other than harddisks, you need to support SCSI commands, it seems like making ATA disk command just be another flavor of scsi in the kernel made sense.
I am not sure how SATA fits in given SAS supports SATA but not the other way around. I believe SATA still uses ATA commands, unless it is not a harddisk in which cases it uses SCSI through ATAPI as it did on IDE.
I have seen suggestions that it makes more sense to think of /dev/sdX as storage disk X rather than scsi disk X. Doesn't help for /dev/sg which is scsi generic, so we probably should just live with it and not worry about the naming as long as it works.
Not like /dev/hdX ever meant IDE. I believe it was used for IDE, MFM, and a few other types that are probably no longer supported.
The RLL modulation type is still hanging around in a couple of different formats for optical drives. Found this out looking for a SATA 6 disc player.
-- Len Sorensen
-- Russell
| From: D. Hugh Redelmeier via talk <talk@gtalug.org> | blkdiscard(8) looks useful. You could apply it to the whole disk before | you start (re-)partitioning. I don't know if you can apply it to a partition. For example, /dev/sda3. The man page does not inform me.
On April 19, 2018 3:29:42 PM CDT, "D. Hugh Redelmeier via talk" <talk@gtalug.org> wrote:
| From: D. Hugh Redelmeier via talk <talk@gtalug.org>
| blkdiscard(8) looks useful. You could apply it to the whole disk before | you start (re-)partitioning.
I don't know if you can apply it to a partition. For example, /dev/sda3. The man page does not inform me.
Apparently mkfs.ext4, with discard enabled will scarify the inodes when the filesystem is created. www.systutorials.com/docs/linux/man/8-mkfs.ext4/ I'm still not sure about weekly trims. That wasn't enabled by systemd by default. I had to do systemctl enable fstrim.timer in order to create the targets. Only got to that last week. I'm kind of a slow ship turning on this build.
--- Talk Mailing List talk@gtalug.org https://gtalug.org/mailman/listinfo/talk
-- Russell
On Fri, Nov 17, 2017 at 09:43:43AM -0500, Russell wrote:
Thanks, I'm sold. I hadn't fully considered forward compatibility in respect of the display. The improved features justify the new price point. Although I'm a little wary of the Realtek S1220A codec. There is kernel support for it in 4.11 and people have made it work. It does have an integrated preamp and pop filter, that works for me
Yeah I guess it gains SPDIF connector, a pair of 10Gbps USB 3.1 gen 2 ports, 2 more SATA connectors internally (I can never have too many of those). Loses the PS/2 port although I sure didn't use that for years myself. Also has a couple of extra audio jacks (Do they have like 8 channel output or something?). I see it also has the header for thunderbolt if you ever want to add a thunderbolt card. The -P didn't support that either. Actually not bad for the extra $44. Looks like a nice machine. Mine are starting to look a bit dated I think. -- Len Sorensen
On November 17, 2017 1:58:04 PM EST, lsorense@csclub.uwaterloo.ca wrote:
Thanks, I'm sold. I hadn't fully considered forward compatibility in respect of the display. The improved features justify the new price
On Fri, Nov 17, 2017 at 09:43:43AM -0500, Russell wrote: point. Although I'm a little wary of the Realtek S1220A codec. There is kernel support for it in 4.11 and people have made it work. It does have an integrated preamp and pop filter, that works for me
Yeah I guess it gains SPDIF connector, a pair of 10Gbps USB 3.1 gen 2 ports, 2 more SATA connectors internally (I can never have too many of those). Loses the PS/2 port although I sure didn't use that for
Given some of the issues with stiching together HID USB endpoint buffers, I considered maintaining that port outside the bottleneck to be a sound policy. I break it now only because I have a better grasp of handling fine grained USB Endpoint descriptors than I did a couple of years ago.
years myself. Also has a couple of extra audio jacks (Do they have like 8 channel output or something?). I see it also has the header for
Yep 8ch / 7.1 surround output. Also impedance sensing output jack's and a task switching mic input. Rather than add an extra output for the .1 subwoofer, you switch the mike to audio out. You wouldn't use a live mic with all that sound, so why not re-task it. They've also placed input and output channels on separate layers and conditioned voltage input. It will be interesting to see if Jack or Pulse audio will add the mic switch to their packages or leave that for the end user.
thunderbolt if you ever want to add a thunderbolt card. The -P didn't support that either. Actually not bad for the extra $44.
I haven't done a detailed comparison feature by feature but at this point, at value, the $44 is not extra. I have done a bit of contracting and every contractor has to build in some wiggle room, but also hit the bid mark. I set my bid mark based on the price I paid for the only preassembled computer I ever bought for myself. Including the external Rockwell modem, capable of 56k, it was a tad over $1100 for everything. (i586 4mg ram etc.) That was 1994 dollars. I set the same value in today's dollars and went from there.
Looks like a nice machine. Mine are starting to look a bit dated I think.
<forrest> Dated is as dated does, as my mama use-ta say. </forrest> It's a feature rich world out there. Sometimes it's hard to see the forrest for the trees. I think we all like to keep hardware costs to a minimum. I sort my priorities into the have to haves and the nice to haves. It was nice to have your help with this. Cheers -- Russell
On Wed, Nov 15, 2017 at 11:51:50AM -0500, Russell via talk wrote:
http://www.coolermaster.com/case/lan-box-haf-series/haf-xb-evo/
What is special about this case? If you stand it on its end, it looks like very thick tower, where harddisks are behind the motherboard. -- William Park <opengeometry@yahoo.ca>
On November 22, 2017 12:01:02 AM EST, William Park via talk <talk@gtalug.org> wrote:
On Wed, Nov 15, 2017 at 11:51:50AM -0500, Russell via talk wrote:
http://www.coolermaster.com/case/lan-box-haf-series/haf-xb-evo/
What is special about this case? If you stand it on its end, it looks like very thick tower, where harddisks are behind the motherboard.
I chose it for the form. I have degeneration in the lower disks of my spine so lift and carry have become issues for me. This form factor is easy to lift and carry. Basically it's an 18 in cube. Stand with your arms extended downward, palms facing in and touching your legs. Raise your arms at the elbows and pick something up. If it is too narrow, this forces your elbows out and puts pressure on your spine; if it is too wide this also forces your elbows away from your body, also causing stress. This is why plastic milk crates are the size they are, to reduce the carry load put on the back of the person carrying them. I don't know what you mean by hard disks being behind the MB but this unit has pretty standardized modular drive cradles. The unit can be opened on the top and sides allowing for ample illumination and access. The planar mount tray is separate from the bottom of the case and about 1/4 the way up from the bottom. This creates a torsion box for added rigidity. You could pick it up by the top rails with the top and sides with the handles removed and not fear over torquing the pcb. The p/s mount is at the bottom which lowers the centre of gravity, also good for moving the unit around. Plus extra filters for air cooling. If the p/s intake fan is on the top, you flip it over, fan down and there is a removable filter tray for dust on the bottom. If the fan is on the side by the plug end, there is an included bracket for extra fans. The whole thing is marketed as optimized for airflow. That alone wouldn't have done it, airflow is usually easily managed, just add more and bigger/better fans. As a total package though, it fit my needs quite nicely.
-- William Park <opengeometry@yahoo.ca> --- Talk Mailing List talk@gtalug.org https://gtalug.org/mailman/listinfo/talk
-- Russell
On Wed, Nov 22, 2017 at 05:59:56AM -0500, Russell via talk wrote:
I chose it for the form. I have degeneration in the lower disks of my spine so lift and carry have become issues for me. This form factor is easy to lift and carry. Basically it's an 18 in cube.
Stand with your arms extended downward, palms facing in and touching your legs. Raise your arms at the elbows and pick something up. If it is too narrow, this forces your elbows out and puts pressure on your spine; if it is too wide this also forces your elbows away from your body, also causing stress. This is why plastic milk crates are the size they are, to reduce the carry load put on the back of the person carrying them.
I don't know what you mean by hard disks being behind the MB but this unit has pretty standardized modular drive cradles. The unit can be opened on the top and sides allowing for ample illumination and access.
The planar mount tray is separate from the bottom of the case and about 1/4 the way up from the bottom. This creates a torsion box for added rigidity. You could pick it up by the top rails with the top and sides with the handles removed and not fear over torquing the pcb.
The p/s mount is at the bottom which lowers the centre of gravity, also good for moving the unit around. Plus extra filters for air cooling. If the p/s intake fan is on the top, you flip it over, fan down and there is a removable filter tray for dust on the bottom. If the fan is on the side by the plug end, there is an included bracket for extra fans. The whole thing is marketed as optimized for airflow. That alone wouldn't have done it, airflow is usually easily managed, just add more and bigger/better fans.
As a total package though, it fit my needs quite nicely.
Looks pretty nice actually. Power supply on the bottom is becoming somewhat common, but the cube design is certainly not as common. Don't have to lie it down to work on the motherboard or cards, which seems nice. -- Len Sorensen
On November 22, 2017 10:47:27 AM EST, lsorense@csclub.uwaterloo.ca wrote:
I chose it for the form. I have degeneration in the lower disks of my spine so lift and carry have become issues for me. This form factor is easy to lift and carry. Basically it's an 18 in cube.
Stand with your arms extended downward, palms facing in and touching your legs. Raise your arms at the elbows and pick something up. If it is too narrow, this forces your elbows out and puts pressure on your spine; if it is too wide this also forces your elbows away from your
On Wed, Nov 22, 2017 at 05:59:56AM -0500, Russell via talk wrote: body, also causing stress. This is why plastic milk crates are the size they are, to reduce the carry load put on the back of the person carrying them.
I don't know what you mean by hard disks being behind the MB but this
unit has pretty standardized modular drive cradles. The unit can be opened on the top and sides allowing for ample illumination and access.
The planar mount tray is separate from the bottom of the case and
about 1/4 the way up from the bottom. This creates a torsion box for added rigidity. You could pick it up by the top rails with the top and sides with the handles removed and not fear over torquing the pcb.
The p/s mount is at the bottom which lowers the centre of gravity,
also good for moving the unit around. Plus extra filters for air cooling. If the p/s intake fan is on the top, you flip it over, fan down and there is a removable filter tray for dust on the bottom. If the fan is on the side by the plug end, there is an included bracket for extra fans. The whole thing is marketed as optimized for airflow. That alone wouldn't have done it, airflow is usually easily managed, just add more and bigger/better fans.
As a total package though, it fit my needs quite nicely.
Looks pretty nice actually. Power supply on the bottom is becoming somewhat common, but the cube design is certainly not as common. Don't have to lie it down to work on the motherboard or cards, which seems nice.
Fits on the upper shelf of the frankenbench with just enough room for the audio amp in the slot underneath. https://www.bhphotovideo.com/c/product/1045658-REG/osd_audio_amp120_2_channe... Nice bridging mode with automatic source switching. When it senses a new signal it switches to that, makes it very versatile. It's kind of vintage so no optical input. I ordered an i7-8700 CPU. @ $487, so I'm $60 shy of my target of $1200 for the build, with no ram yet. The ram will put me 10% over estimate. Rather than monkey with the CPU down the road, I'll add in the water-cooling as well. I'll do this after I seat the parts and see what's what. So probably 15-17% over for the feature improvements and lesser physical storage on board, which is not a real issue at this time. Looks good on paper to me. We'll see how the build goes. I was never able to work on a car without scraping my knuckles, system integration is kind of like that too. Thanks for the help.
-- Len Sorensen
-- Russell
Top posting this as the hardware build is complete and functioning. Case = CoolerMaster HAF XB EVO - easy to work in, on and around. Added a Masterliquid 240 water-cooling unit. The rad took up the case mounting holes at the front, so I jury rigged the case fans as pull units. It's called duct tape for a reason :-) I'll cobble up an offset adapter for the original case fans later. Quiet, you bet. When I fired it up, I had to put my hand on the w/c pump to see if it was really on. 4 fans push/pull @ 850 rpm on the rad keep the temp around 30 c. although there's no real load as of yet. I can't even hear the fans if I listen for them. I have to speed them up manually in bios to hear what they sound like. CPU = i7-8700 - locked so no overclocking Asus Prime z370-A - You can get the board today at Canada Computers for $174.00 after 20$ mail in rebate. I had no problem with the audio codec. It worked right off. Intel UHD graphics seems to have lost DRM support in Fedora 27. It was apparently ok in 25, so something should be available down the road. Had to beef up display graphics capabilities by adding i915.alpha_support=1 to boot. Have to figure out efi & grub2 to make it persistent. RAM = 1 x 8 DDR4 2666 SSD = Intel 512 Gig - over-provisioned it shows up as 475 Gig. I left an additional 40 Gig undefined. Separate partitions for / (40Gig)& /boot/efi (40Gig) & /swap (16Gig) & /home I also had to pick up a 500w p/s. My 6+2 power dongle on the existing one showed up on the diagnostic LED's as a CPU fault. I even ordered an RMA label on the CPU before I figured it out. Jumped the gun because of fear of Christmas congestion. Thanks Lennart for the upgrade MB suggestion, I thought it was good value at retail. I bought another one on sale yesterday, when I went downtown and got the displayport cable and PSU. Also thanks Hugh for the SSD wear balancing primer and other suggestions. True to my word about skinned knuckles, I cut my thumb on the cheap aluminum case insert plate for the external IO. Fedora boots in 10 sec and journalctl just shows the dbus PCI error which I assume is related to DRM graphics. Other stuff. I couldn't install the CPU upside-down because the depression for the fan is too far back. Probably meant for a much larger supply. However, I just noticed thumbscrews in an odd location by the PSU and perhaps there is a built in offset for a smaller unit to be able to use the intake filter. Just have to reverse it up. I haven't totalled receipts yet but I'm within 15% over of my original budget, without a HDD, but with extra features that made to all worthwhile for me. Cheers Russell On November 22, 2017 11:54:03 AM EST, Russell <rreiter91@gmail.com> wrote:
I chose it for the form. I have degeneration in the lower disks of my spine so lift and carry have become issues for me. This form factor is easy to lift and carry. Basically it's an 18 in cube.
Stand with your arms extended downward, palms facing in and touching your legs. Raise your arms at the elbows and pick something up. If it is too narrow, this forces your elbows out and puts pressure on your spine; if it is too wide this also forces your elbows away from your
On Wed, Nov 22, 2017 at 05:59:56AM -0500, Russell via talk wrote: body, also causing stress. This is why plastic milk crates are the size they are, to reduce the carry load put on the back of the person carrying them.
I don't know what you mean by hard disks being behind the MB but
On November 22, 2017 10:47:27 AM EST, lsorense@csclub.uwaterloo.ca wrote: this
unit has pretty standardized modular drive cradles. The unit can be opened on the top and sides allowing for ample illumination and access.
The planar mount tray is separate from the bottom of the case and
about 1/4 the way up from the bottom. This creates a torsion box for added rigidity. You could pick it up by the top rails with the top and sides with the handles removed and not fear over torquing the pcb.
The p/s mount is at the bottom which lowers the centre of gravity,
also good for moving the unit around. Plus extra filters for air cooling. If the p/s intake fan is on the top, you flip it over, fan down and there is a removable filter tray for dust on the bottom. If the fan is on the side by the plug end, there is an included bracket for extra fans. The whole thing is marketed as optimized for airflow. That alone wouldn't have done it, airflow is usually easily managed, just add more and bigger/better fans.
As a total package though, it fit my needs quite nicely.
Looks pretty nice actually. Power supply on the bottom is becoming somewhat common, but the cube design is certainly not as common. Don't have to lie it down to work on the motherboard or cards, which seems nice.
Fits on the upper shelf of the frankenbench with just enough room for the audio amp in the slot underneath.
https://www.bhphotovideo.com/c/product/1045658-REG/osd_audio_amp120_2_channe...
Nice bridging mode with automatic source switching. When it senses a new signal it switches to that, makes it very versatile. It's kind of vintage so no optical input.
I ordered an i7-8700 CPU. @ $487, so I'm $60 shy of my target of $1200 for the build, with no ram yet. The ram will put me 10% over estimate. Rather than monkey with the CPU down the road, I'll add in the water-cooling as well. I'll do this after I seat the parts and see what's what. So probably 15-17% over for the feature improvements and lesser physical storage on board, which is not a real issue at this time.
Looks good on paper to me. We'll see how the build goes.
I was never able to work on a car without scraping my knuckles, system integration is kind of like that too.
Thanks for the help.
-- Len Sorensen
-- Russell
-- Russell
On Mon, Nov 27, 2017 at 02:21:33PM -0500, Russell wrote:
Top posting this as the hardware build is complete and functioning.
Case = CoolerMaster HAF XB EVO - easy to work in, on and around. Added a Masterliquid 240 water-cooling unit. The rad took up the case mounting holes at the front, so I jury rigged the case fans as pull units. It's called duct tape for a reason :-) I'll cobble up an offset adapter for the original case fans later.
Quiet, you bet. When I fired it up, I had to put my hand on the w/c pump to see if it was really on. 4 fans push/pull @ 850 rpm on the rad keep the temp around 30 c. although there's no real load as of yet. I can't even hear the fans if I listen for them. I have to speed them up manually in bios to hear what they sound like.
CPU = i7-8700 - locked so no overclocking
Asus Prime z370-A - You can get the board today at Canada Computers for $174.00 after 20$ mail in rebate. I had no problem with the audio codec. It worked right off. Intel UHD graphics seems to have lost DRM support in Fedora 27. It was apparently ok in 25, so something should be available down the road. Had to beef up display graphics capabilities by adding i915.alpha_support=1 to boot. Have to figure out efi & grub2 to make it persistent.
RAM = 1 x 8 DDR4 2666
SSD = Intel 512 Gig - over-provisioned it shows up as 475 Gig. I left an additional 40 Gig undefined. Separate partitions for / (40Gig)& /boot/efi (40Gig) & /swap (16Gig) & /home
I also had to pick up a 500w p/s. My 6+2 power dongle on the existing one showed up on the diagnostic LED's as a CPU fault. I even ordered an RMA label on the CPU before I figured it out. Jumped the gun because of fear of Christmas congestion.
Oh yes a 6+2 is for graphics, not the same as a 4+4 for CPU. Totally different pinout (almost reversed in fact). I am surprised if they fit in each other, but I have never tried. -- Len Sorensen
On November 27, 2017 5:14:42 PM EST, lsorense@csclub.uwaterloo.ca wrote:
Top posting this as the hardware build is complete and functioning.
Case = CoolerMaster HAF XB EVO - easy to work in, on and around. Added a Masterliquid 240 water-cooling unit. The rad took up the case mounting holes at the front, so I jury rigged the case fans as pull units. It's called duct tape for a reason :-) I'll cobble up an offset adapter for the original case fans later.
Quiet, you bet. When I fired it up, I had to put my hand on the w/c
On Mon, Nov 27, 2017 at 02:21:33PM -0500, Russell wrote: pump to see if it was really on. 4 fans push/pull @ 850 rpm on the rad keep the temp around 30 c. although there's no real load as of yet. I can't even hear the fans if I listen for them. I have to speed them up manually in bios to hear what they sound like.
CPU = i7-8700 - locked so no overclocking
Asus Prime z370-A - You can get the board today at Canada Computers
for $174.00 after 20$ mail in rebate.
I had no problem with the audio codec. It worked right off. Intel UHD graphics seems to have lost DRM support in Fedora 27. It was apparently ok in 25, so something should be available down the road. Had to beef up display graphics capabilities by adding i915.alpha_support=1 to boot. Have to figure out efi & grub2 to make it persistent.
RAM = 1 x 8 DDR4 2666
SSD = Intel 512 Gig - over-provisioned it shows up as 475 Gig. I left an additional 40 Gig undefined. Separate partitions for / (40Gig)& /boot/efi (40Gig) & /swap (16Gig) & /home
I also had to pick up a 500w p/s. My 6+2 power dongle on the existing one showed up on the diagnostic LED's as a CPU fault. I even ordered an RMA label on the CPU before I figured it out. Jumped the gun because of fear of Christmas congestion.
Oh yes a 6+2 is for graphics, not the same as a 4+4 for CPU. Totally different pinout (almost reversed in fact).
I am surprised if they fit in each other, but I have never tried.
Sorry I misspoke. I tried the 6x2 as a last effort. You're right, there's no way to make it fit. It was a 4x4 snapped together, one plug in constant use. I guess the other was a dud or perhaps it was corroded and threw off spikes.
-- Len Sorensen
-- Russell
participants (5)
-
D. Hugh Redelmeier -
Jamon Camisso -
lsorense＠csclub.uwaterloo.ca -
Russell -
William Park