Intel i7-6700 vs Xeon E3-1230 v5 --> 65W vs 80W ??
Intel i7-6700 and Xeon E3-1230 v5 are pretty much the same, except that i7 has builtin Graphics and Xeon does not. I don't get it. How can i7 with builtin graphics consume less power?-- William
32 vs 14 nm lithography probably do it? -tl On Tue, Jun 28, 2016 at 10:52 PM, William Park via talk <talk@gtalug.org> wrote:
Intel i7-6700 and Xeon E3-1230 v5 are pretty much the same, except that i7 has builtin Graphics and Xeon does not. I don't get it. How can i7 with builtin graphics consume less power? -- William
--- Talk Mailing List talk@gtalug.org https://gtalug.org/mailman/listinfo/talk
They are both 14nm. Xeon turbos to 3.8GHz and no GPU at 80W, but i7 turbos to 4GHz with GPU at 65W. Me thinks it's Intel marketing. Take dies that fail QA, rebrand them as "Xeon", and sell just below i7. -- William On Wed, Jun 29, 2016 at 12:49:50AM -0400, ted leslie wrote:
32 vs 14 nm lithography probably do it?
-tl
On Tue, Jun 28, 2016 at 10:52 PM, William Park via talk <talk@gtalug.org> wrote:
Intel i7-6700 and Xeon E3-1230 v5 are pretty much the same, except that i7 has builtin Graphics and Xeon does not. I don't get it. How can i7 with builtin graphics consume less power? -- William
--- Talk Mailing List talk@gtalug.org https://gtalug.org/mailman/listinfo/talk
Top posting makes this confusing. I tried to fix this. On Tue, Jun 28, 2016 at 10:52 PM, William Park via talk <talk@gtalug.org> wrote: | | Intel i7-6700 and Xeon E3-1230 v5 are pretty much the same, except that i7 | has builtin Graphics and Xeon does not. I don't get it. How can i7 with | builtin graphics consume less power? | From: William Park via talk <talk@gtalug.org> | Date: Fri, 1 Jul 2016 12:30:56 -0400 | | They are both 14nm. Xeon turbos to 3.8GHz and no GPU at 80W, but i7 | turbos to 4GHz with GPU at 65W. Me thinks it's Intel marketing. Take | dies that fail QA, rebrand them as "Xeon", and sell just below i7. Good points. Here are my guesses (I have no actual knowledge). Here's Intel'S diff: <http://ark.intel.com/compare/88196,88182> Notice on the first line "Product name", the turbo speed is given for the i7 but the base frequency is given for the Xeon. Different spins for different markets? The i7 does not support ECC but the Xeon does, putting them in separate markets. The increased TDP is puzzling until you understand that CPUs now throttle when parts of them get hot. The Xeon has more headroom so (with the correct cooling) it probably runs faster under continous loads. Perhaps if you gave the i7 80w cooling (plus some for the GPU) it might match the Xeon. The faster turbo might get in the way. Lots of folks are surprised to find Intel CPUs don't run as fast as they are specified to. The blame is almost always on the cooling. But it may not be possible to cool the processors in question sufficiently. For example, many fanless Cherrytrail systems don't decode video the at the rate their specs suggest. I've seen several folks claim from experience that the Kangaroo is unsuitable for being a TV box for this reason. <http://www.tomshardware.com/answers/id-3021966/cooling-infocus-kangaroo.html>
On Fri, Jul 01, 2016 at 12:30:56PM -0400, William Park via talk wrote:
They are both 14nm. Xeon turbos to 3.8GHz and no GPU at 80W, but i7 turbos to 4GHz with GPU at 65W. Me thinks it's Intel marketing. Take dies that fail QA, rebrand them as "Xeon", and sell just below i7.
That is exactly what they are doing. The dies that work at low voltage get sold as i7 with a TDP of 65W, while the ones that need more voltage or have a broken GPU get the GPU disabled and are sold as the low end Xeon. So you might get a Xeon that runs at 65W but is a xeon simply because the GPU was broken, or you might get one with a perfectly working GPU but which needed a bit more than 65W to do the specs. Either way you get what intel promised. You might get better than that, but no promises from intel. -- Len Sorensen
On 07/04/2016 11:51 AM, Lennart Sorensen via talk wrote:
They are both 14nm. Xeon turbos to 3.8GHz and no GPU at 80W, but i7 turbos to 4GHz with GPU at 65W. Me thinks it's Intel marketing. Take dies that fail QA, rebrand them as "Xeon", and sell just below i7. That is exactly what they are doing. The dies that work at low voltage get sold as i7 with a TDP of 65W, while the ones that need more voltage or have a broken GPU get the GPU disabled and are sold as the low end Xeon. So you might get a Xeon that runs at 65W but is a xeon simply because
On Fri, Jul 01, 2016 at 12:30:56PM -0400, William Park via talk wrote: the GPU was broken, or you might get one with a perfectly working GPU but which needed a bit more than 65W to do the specs. Either way you get what intel promised. You might get better than that, but no promises from intel.
It would be kind of scary to think that the Intel server processors are just failed consumer products. Does that mean that they are also turning off the ECC in the I7? What about the ECC in the cache( I know some processors carry ECC through the cache but not sure about Intel )? -- Alvin Starr || voice: (905)513-7688 Netvel Inc. || Cell: (416)806-0133 alvin@netvel.net ||
On Mon, Jul 04, 2016 at 02:51:57PM -0400, Alvin Starr via talk wrote:
It would be kind of scary to think that the Intel server processors are just failed consumer products.
Well the dies have variations. They have chosen a different requirement for the i7 than the E3. Servers usually have better cooling design, so 80W is no big deal. It is just as likely a chip with a broken ECC memory controller could end up as a lower end i7 or even i5 chip.
Does that mean that they are also turning off the ECC in the I7? What about the ECC in the cache( I know some processors carry ECC through the cache but not sure about Intel )?
Yes they would be just turning off ECC support. Internally I am pretty sure even the i7 runs with ECC on the cache. -- Len Sorensen
participants (5)
-
Alvin Starr -
D. Hugh Redelmeier -
Lennart Sorensen -
ted leslie -
William Park