News Intel Brings Its Own Benchmark to Refute AMD's '2X' EPYC Claim

[PaulAlcorn]

Intel fires back at AMD CEO Lisa Su's Computex claims that the company's upcoming 7nm EPYC Rome data center processors are twice as fast.

Intel Brings Its Own Benchmark to Refute AMD's '2X' EPYC Claim : Read more

 

[amir.zabin]

the intel 56Core dosent have hyperthreading, thats confirmed now, and by what i see so far amd can optimize the compiler as well so intel is rekt tbh

 

[jimmysmitty]

the intel 56Core dosent have hyperthreading, thats confirmed now, and by what i see so far amd can optimize the compiler as well so intel is rekt tbh

That is incorrect per reporting:

https://www.anandtech.com/show/14182/hands-on-with-the-56core-xeon-platinum-9200-cpu-intels-biggest-cpu-package-ever

It does have SMT on it. Why would Intel not use SMT on their products?

There is also more to server than just core count as well. The technology around a CPU matters greatly and Intel typically does have more of that especially since they typically have a hand in developing newer standards.

AMD will probably grab some market share but it is not going to be an overnight change from Intel to AMD.

However this goes to show that you can never trust marketing from a company. They will always show their product in the best light possible. AMD and Intel both do it as do most other companies. But it is to be expected. You want to sell your product so you claim its the best.

 

[bigdragon]

Intel needs to start specifying if they applied Spectre, Meltdown, MDS, and other mitigations to their benchmark machines and scores. The difference can easily reach into 40% in some cases. Intel's response makes me think AMD applied the mitigations when calculating their figures while Intel removed the mitigations in their response numbers.

 

[EvilOni]

The results are scarcely comparable.

those Xeon 9000 series practically don't exist. You can't even find pricing on them, it's only available through select OEMs at an "if you have to ask for the price you probably can't afford it" basis.

The CPU's that AMD compared their EPYCS to were $10k a pop and the 9000's are more expensive at that.

And each have what? a 400w TDP?

Yeah a CPU that's four times more expensive with twice the TDP to only draw even is going to be a hot seller in the datacenter

 

[cement_head]

We just bought current generation AMD EPYC CPU's - a dual socket - for bioinformatics work. Why did we go with AMD rather than Intel? One, we've been using an AMD Opteron CPU for several years and it's performance has been excellent, and two, for the current EPYC models/choices, price was a major factor. EPYC CPUs are significantly cheaper - more than 50% less than a comparable (performance-wise) Intel CPU, so we can spend more money on RAM and GPUs. With AMD's EPYC line of CPUs, Intel's costs just don't make sense...

 

[amir.zabin]

That is incorrect per reporting:

https://www.anandtech.com/show/14182/hands-on-with-the-56core-xeon-platinum-9200-cpu-intels-biggest-cpu-package-ever

It does have SMT on it. Why would Intel not use SMT on their products?

There is also more to server than just core count as well. The technology around a CPU matters greatly and Intel typically does have more of that especially since they typically have a hand in developing newer standards.

AMD will probably grab some market share but it is not going to be an overnight change from Intel to AMD.

However this goes to show that you can never trust marketing from a company. They will always show their product in the best light possible. AMD and Intel both do it as do most other companies. But it is to be expected. You want to sell your product so you claim its the best.

according to the intel slide "hyperthreading:ON 1thread per core"
dont forget that they dont have the software updates installed in their benchmarks just some food for thought

 

[JayNor]

The Cooper Lake 14nm 48 core/96 thread chip will add two more memory channels and support for higher speed memory. Also will add bfloat16 avx512 support. wikichip shows it with pcie4 support, although I haven't seen an intel announcement yet. This is supposed to be sampling this year, so need to run the benchmarks again to see if the extra features boost it over the amd 64 core chip.

 

[JayNor]

That is incorrect per reporting:

https://www.anandtech.com/show/14182/hands-on-with-the-56core-xeon-platinum-9200-cpu-intels-biggest-cpu-package-ever

It does have SMT on it. Why would Intel not use SMT on their products?

The avx512 is shared between threads. Apps that use it are trying to keep it filled. I've read ai performance recommendations that say turn off hyperthreading and use core affinity assignments to guarantee consistent high performance from these.

 

[Ncogneto]

You call that firing back? LOL, more like a wimper if you ask me. Use the $ spent on the system that Intel used to pull even with AMD, build the best system spending the same amount of money on an AMD system, then benchmark the two head to head. That would be a fair benchmark. And none of this special Kernel compiled to run cinebench.

 

[Julian321]

No software patches installed it seems. That could easily knock at least 12% performance. Regardless, in terms of quality it's not the performance that really matters, its the spread between the TDP and the performance. Intels 9000 series servers may have more raw power but it consumes too much electricity and requires too much cooling. It's a sloppy design and even if they were the same price I would chose the AMD system.

 

[EpicAnchor]

I feel Intel has been cheating for a long time. Only recently have they been caught. They design their chips for maximum performance, without security protection in mind. Because, this would reduce performance. Intel systems with patches installed are closer to how they should have been designed in the first place. Even though AMD chips are more secure without any patches, than Intel's with patches. Probably why Intel's new 10nm chips are slower than current chips at the same speed. I have seen a lot of Intel's shenanigans over the last 30 years!

 

[redgarl]

SO AMD forgot to include TEH OPTIMIZATION patches... AMD also didn't include TEH VULNERABILITY patches....

Intel didn't disable HT for that test huh...

 

[mihen]

Obviously an Apples to Potatoes comparison. Still there is a selling point for Intel until they move into 10nm. Most software is built around their CPUs. In professional applications its all about how the hardware performs on specific software and they will be testing combinations before a hard commitment.

 

[redgarl]

No software patches installed it seems. That could easily knock at least 12% performance. Regardless, in terms of quality it's not the performance that really matters, its the spread between the TDP and the performance. Intels 9000 series servers may have more raw power but it consumes too much electricity and requires too much cooling. It's a sloppy design and even if they were the same price I would chose the AMD system.

Prescott anyone?

 

[redgarl]

Obviously an Apples to Potatoes comparison. Still there is a selling point for Intel until they move into 10nm. Most software is built around their CPUs. In professional applications its all about how the hardware performs on specific software and they will be testing combinations before a hard commitment.

They are obviously doing some Damage Control. They know it is going to be a long time before 7nm get there.

 

[joeblowsmynose]

"Your processors aren't over twice as fast as ours! They're only 1.8x faster with our optimizations applied!" -- Intel

 

[Pantszilla]

Intel fires back at AMD CEO Lisa Su's Computex claims that the company's upcoming 7nm EPYC Rome data center processors are twice as fast.

Intel Brings Its Own Benchmark to Refute AMD's '2X' EPYC Claim : Read more

HAHAHAHAHAHA "But Intel says that AMD didn't configure its Intel test system correctly" like those tests Intel did where they disabled HT on Ryzen... or that 5GHz 16 core on a custom 1HP liquid cooler refrigerator?

 

[MasterMadBones]

The key here is AVX-512, which we know Rome will not support. However, the number of use cases fot this instruction set is still small, and in any other instruction set, AMD would lead.

Also, apparently Intel's AVX-512 implementation performs best with HTT disabled. In realistic scenarios, such as compute-for-rent, this is simply unacceptable.

 

[jimmysmitty]

I feel Intel has been cheating for a long time. Only recently have they been caught. They design their chips for maximum performance, without security protection in mind. Because, this would reduce performance. Intel systems with patches installed are closer to how they should have been designed in the first place. Even though AMD chips are more secure without any patches, than Intel's with patches. Probably why Intel's new 10nm chips are slower than current chips at the same speed. I have seen a lot of Intel's shenanigans over the last 30 years!

Except AMD has not had 99% of the server market nor the majority of the desktop market for the past 10 years. Intel is a bigger target thus more vulnerabilities.

As well Intel is not the only one with the flaws. While Intel does have more AMD has been affected as well as has VIA.

The more share AMD gets, the larger they get the more targeted they will be and the more we will see coming for them as well.

No software patches installed it seems. That could easily knock at least 12% performance. Regardless, in terms of quality it's not the performance that really matters, its the spread between the TDP and the performance. Intels 9000 series servers may have more raw power but it consumes too much electricity and requires too much cooling. It's a sloppy design and even if they were the same price I would chose the AMD system.

I have not seen anything stating it does or does not have patching. However it looks like their tests were done on Linux which has been affected differently than Windows.

 

[Julian321]

Except AMD has not had 99% of the server market nor the majority of the desktop market for the past 10 years. Intel is a bigger target thus more vulnerabilities.

As well Intel is not the only one with the flaws. While Intel does have more AMD has been affected as well as has VIA.

The more share AMD gets, the larger they get the more targeted they will be and the more we will see coming for them as well.



I have not seen anything stating it does or does not have patching. However it looks like their tests were done on Linux which has been affected differently than Windows.

See figure one in the paragraph. "Performance results are based on testing...and may not reflect all publicly available security results...". So intel, you cannot definitely tell me that you are using all of the security patches, both publicly and privately available in this benchmark? So yeah, they haven't used all/some of the security updates. This benchmark is little more than propaganda.

 

[bit_user]

note, this Linux kernel is patched for the most recent vulnerabilities

The Intel kernel? According to whom?

What distro did they use - Intel's own Clear Linux, perhaps?

 

[bit_user]

However this goes to show that you can never trust marketing from a company. They will always show their product in the best light possible. AMD and Intel both do it as do most other companies. But it is to be expected. You want to sell your product so you claim its the best.

Really? If AMD really wanted to skew the results, why did they go to the trouble of using Intel's own compiler, icc?

I think AMD probably took the approach of configuring the Intel system according to standard practice. Intel clearly optimized their configuration to a degree well beyond what the typical user would do. I don't think there's any evidence of them actively sabotaging the Intel data, even if they didn't bend over backwards to optimize it to the hilt as Intel clearly did.

So, is AMD's benchmark really invalid? Depends on who the user is and how motivated they are to extract the maximum performance from their hardware. For most people, I think the AMD results are legit.

What's most troubling is your conclusion that just because Intel can find a way to squeeze out more performance is somehow evidence that AMD was acting in bad faith. I sincerely doubt that anyone can optimize software for Intel's CPUs as well as Intel can, so probably no one will ever produce results as good as Intel's.

 

[bit_user]

BTW, what I really want to know is whether Intel's configuration passes NAMD's regression tests. If their optimizations sacrifice numerical accuracy to any significant degree, their results should be considered invalid.

 

[bit_user]

The results are scarcely comparable.

those Xeon 9000 series practically don't exist. You can't even find pricing on them, it's only available through select OEMs at an "if you have to ask for the price you probably can't afford it" basis.

The way to address this would be to specify the as-configured price. Unfortunately, because AMD's CPU is currently unreleased, they can't necessarily offer such pricing data.