Intel Xeon Platinum 8176 Scalable Processor Review

[TJ Hooker]

Do these CPUs have the same thermal issues as the i9 series?

I know these aren't going to be overclocked, but the additional CPU temps introduce a number of non-trivial engineering challenges that would result in significant reliability issues if not taken into account.

Specifically, as thermal resistance to the heatsink increases, the thermal resistance to the motherboard drops with the larger socket and more pins. This means more heat will be dumped into the motherboard's traces. That could raise the temperatures of surrounding components to a point that reliability is compromised. This is the case with the Core i9 CPUs.

See the comments here for the numbers:
http://www.tomshardware.com/forum/id-3464475/skylake-mess-explored-thermal-paste-runaway-power.html

You mean thermal issues that will never be seen because server CPUs are never OCed? Most server CPUs will not be maxed out 24x7. A single server with this CPU will probably be cut up into at least 6 different server roles using VM.

Either way the i9 seems to be fine at stock speeds. The biggest issues arise when overclocking, which is the same with every CPU.

In Tom's review they found that they required a high end 240 mm AIO or custom loop to cool an i9-7900X under heavy load even at stock. From their follow up article: "Skylake-X at its stock settings can barely be cooled during normal operation."

 

[daglesj]

Will look forward to picking one up on Ebay for £50 in a few years.

 

[kingbob2]

so I'm familiar with server processors and the high core count vs clock speed factoring in for the high prices, but can anyone explain why a 4C/8T processor is going for over $7k? https://ark.intel.com/products/120499/Intel-Xeon-Platinum-8156-Processor-16_5M-Cache-3_60-GHz

 

[aldaia]

On a side note Anandtech did a limited review of the $4000 EPYC 7601 against $8000 Xeon 8176. It's a very limited comparison, but EPYC is mostly beating the xeon, specially on FP benchmarks, while using less power (despite it's higher TDP).

Honestly I didn't expect EPYC to do so well in FP. A pity that EPYC missed recent January top 500, I expect to see some entries though in the coming November top 500.

http://www.anandtech.com/show/11544/intel-skylake-ep-vs-amd-epyc-7000-cpu-battle-of-the-decade

 

[bit_user]

so I'm familiar with server processors and the high core count vs clock speed factoring in for the high prices, but can anyone explain why a 4C/8T processor is going for over $7k? https://ark.intel.com/products/120499/Intel-Xeon-Platinum-8156-Processor-16_5M-Cache-3_60-GHz

Although I'm not sure they've sold one for quite that much, you can find very expensive quad-core E5 and E7 Xeons through many prior generations.

The reason is probably explained by some combination of the following:

• 
  ■ UPI Links: 3
  ■ Memory Channels: 6
  ■ Max Memory: 768 GB
  ■ PCIe 3.0 Lanes: 48
  ■ ECC Memory: yes

So, basically, it's for customers with workloads demanding one or more of the following:

• 
  ■ large amounts of memory (with 3 UPI links, you can have up to 8x of these for 6 TB of RAM)
  ■ a large amount of I/O (e.g. networking and NVMe)
  ■ PCIe-connected GPU computing accelerators (again multiply that 48 PCIe 3.0 lanes by up to 8 CPUs...)
  ■ RAS (Reliability And Stability) features offered by that platform chipset & BIOS.

 

[bit_user]

On a side note Anandtech did a limited review of the $4000 EPYC 7601 against $8000 Xeon 8176. It's a very limited comparison, but EPYC is mostly beating the xeon, specially on FP benchmarks, while using less power (despite it's higher TDP).

Honestly I didn't expect EPYC to do so well in FP. A pity that EPYC missed recent January top 500, I expect to see some entries though in the coming November top 500.

EPYC is only strong in non-AVX optimized FP. Their AVX implementation is weak and they don't do AVX-512 at all. AMD is banking on AVX-intensive workloads instead being accelerated on GPUs (which is correct, in many cases).

I actually found their integer performance a bit more interesting, since it's largely devoid of such caveats. I honestly didn't expect it to do quite so well, there. However, these results should be taken with a grain of salt, as the author notes:

Ok, first a disclaimer. The SPECint rate test is likely unrealistic. If you start up 88 to 128 instances, you create a massive bandwidth bottleneck and a consistent CPU load of 100%, neither of which are very realistic in most integer applications. You have no synchronization going on, so this is really the ideal case for a processor such as the AMD EPYC 7601. The rate test estimates more or less the peak integer crunching power available, ignoring many subtle scaling problems that most integer applications have.

 

[aldaia]

On a side note Anandtech did a limited review of the $4000 EPYC 7601 against $8000 Xeon 8176. It's a very limited comparison, but EPYC is mostly beating the xeon, specially on FP benchmarks, while using less power (despite it's higher TDP).

Honestly I didn't expect EPYC to do so well in FP. A pity that EPYC missed recent January top 500, I expect to see some entries though in the coming November top 500.

EPYC is only strong in non-AVX optimized FP. Their AVX implementation is weak and they don't do AVX-512 at all. AMD is banking on AVX-intensive workloads instead being accelerated on GPUs (which is correct, in many cases).

Exactly, that is why I'm surprised that they are doing so well in FP benchmarks. I Expected EPYC to trail the xeons here. Especially considering that:

the NAMD binary is compiled with Intel ICC and optimized for AVX.