In Depth Processor Comparison: Ivy Bridge E5-2673 v2 vs E5-2667 v2

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

Storm-Chaser

Twin Turbo
Apr 16, 2020
151
25
28
Upstate NY
There has been talk in one of the other sub forums:

HP z820 Build Thread

That the 110 watt TDP OEM E5-2673 v2 CPU will have reduced performance when compared to its retail equivalent, the 130 watt TDP E5-2667 v2 processor, due to having a "lower TDP". Yes, some people are insisting that this will limit performance and despite sharing identical frequency clocking and turbo specifications, the OEM 110w chip will be outperformed by it's 130 watt TDP "big brother" retail chip because it has a higher TDP. The 110w E5-2673 v2 is a rare processor that is difficult to find in the US market, as it was OEM only and never offered for retail purchase.

On paper, these processors are identical in clock speed and identical in terms of turbo configuration, each with a single core turbo speed of 4.0GHz and an all core turbo speed of 3.6GHz.


These are nearly identical LGA 2011 Ivy Bridge chips with 8 cores and 16 threads apiece. Despite showing real world benchmarks, people are still insisting that the OEM chip is slower. So to settle this score once and for all, I have ordered two E5-2667 v2 chips for comparison with my two existing E5-2673 v2 chips. We need a real world, toe to toe identical hardware comparison and that's exactly what I'm setting out to do in this thread. And find out once and for all if there is a performance deviation between the two chips.

First of all, it will be another week or two before the new 130 watt E5-2667 v2 processors arrive, so in the mean time I wanted to get prepared with my hardware/bench testing rigs for this comparison, since this benchmarking routine will be quite extensive. Second, I am testing these processors in a dual socket environment. Meaning two processors vs two processors.

I have two test rigs ready to go in which performance will be measured and benchmarks will be performed. Meaning identical hardware comparisons, with identical bios settings.

The first is a Dell PowerEdge R620 server.
This machine has 96GB of DDR3 (24 slots) memory running at 1333 MHz and all switches have been flipped in the BIOS for maximum performance, disabling all power saving modes, and including activating full turbo mode and running 8 channels of memory bandwidth. Second machine is a dual socket workstation of the same vintage, an HP z820. This machine is roughly similar in terms of specs relative to the PowerEdge r620, also having eight channels of memory performance (16 slots), but running at 1866MHz, and will offer a slightly different environment for testing the chips, giving us more data on real world performance numbers from these CPUs across diverse hardware. The z820 BIOS settings will be set to full performance modes for both memory and CPU. All CPU power saving modes will be disabled.

We will be benchmarking from AIDA64 first, including the full benchmark report for a very clear and concise performance comparison. I will also include Cinebench R15, R15 extreme, and R20 as well. Also, we will be using other benchmarks as well, such as the quick CPU-z bench, AIDA64 Cache and memory benchmark and WPrime 1024m and perhaps a few others. Stay tuned for results from the E5-2673 v2 first and once the retail processors arrive we can draw some "solid" performance conclusions here. And see if these OEM processors have any "advantages" over their retail counterparts.

In the meantime, here are the specs on the processors we will be working with, courtesy of CPU World:
What do you think? Will the higher TDP processor come out on top? Vote now!

1601100851976.png
1601101092888.png
1601101158076.png

Individual Chip specifications:
1601101376173.png

VERSUS

1601101431960.png
 

Attachments

Last edited:

Storm-Chaser

Twin Turbo
Apr 16, 2020
151
25
28
Upstate NY
Dell PowerEdge R620
AIDA64 Full Benchmark Report -
2 x Xeon E5-2673 v2 Processors
96GB DDR3 @ 1333MHz (24 Modules)

PLEASE NOTE:
AIDA64 fails to see a difference between the E5-2667 v2 and the E5-2673 v2, consequently, the report below erroneously reports I have two E5-2667 v2 processors, but I can assure you, they are the OEM E5-2673 v2. Just so you know.
 

Attachments

  • Like
Reactions: Aluminat

Storm-Chaser

Twin Turbo
Apr 16, 2020
151
25
28
Upstate NY
the tr 2990wx with photoworxx looks terrible.

u could also try intel linpack:


the large matrix cases are of interest.
--
Yes now that you mention it, the 3960X is also lagging behind - and should likely beat my dual socket Xeon system, right? I'll be updating AIDA64 for starters, perhaps we can draw more accurate comparisons with up to date scoring.

Stay tuned for AIDA64 performance results from the HP z820 workstation, as I am prepping the system now. Will be downloading your benchmark recommendation as well.
 

Storm-Chaser

Twin Turbo
Apr 16, 2020
151
25
28
Upstate NY
Free mulligan, right now, if you want it... I highly suggest you take the offer before reading below.

Due to some unexpected covid stuff and helping my parents out, I have not been able to test the new E5 2667 v2 processor but I want to be clear it will basically perform identically to the 2667 v2, because they have identical turbo configurations and identical frequency multipliers. Have you guys not heard of undervolting? I can prove I am dead on the money with my theory about identical performance and I can prove it.

Go to any rig that will allow you to modify the CPU voltage. Start with a voltage number on the high side of the spectrum and run some CPU benchmarks. Save your results and then adjust the CPU voltage down to the low side of the spectrum. Now boot it at the lower vcore and run the same CPU benchmarks again.

Good read on exactly why can be found here:

.