AMD EPYC 9754 Benchmarks For The 128-Core Bergamo

[phoronix]

Phoronix: AMD EPYC 9754 Benchmarks For The 128-Core Bergamo

In addition to the review embargo lift today for Genoa-X with our AMD EPYC 9684X benchmarks, the lift is also today on the new AMD EPYC "Bergamo" processors for offering up to 128 cores / 256 threads per socket using the new Zen 4C core. In this article is an initial look at the performance provided by the AMD EPYC 9754 128-core processors.

https://www.phoronix.com/review/amd-epyc-9754-bergamo

[CochainComplex]

It feels like ...two more gens and intel will drop entirely from the list

[schmidtbag]

Strange how the 8490H has a MSRP that is $5000 higher than the 9754...

[NeoMorpheus]

It feels like ...two more gens and intel will drop entirely from the list

A man can only dream...

Meanwhile, my dream goes like this:

[sophisticles]

Two things:

1) To those that think it would be great of Intel somehow either ceased to exist to ceased to be competitive, what do you think would happen to AMD.
There's an old adage that competition improves the breed, we should be hoping that Intel's E-core only Xeons are beasts that take the performance crown, so that AMD will be forced to improve their products again.

2) Michael, why not include that beastly Xeon Max based server you have, with all that onboard ram I'm sure Intel would have taken a few more wins.

I'm guessing maybe you are precluded from doing so by NDA.

[Michael]

Two things:

1) To those that think it would be great of Intel somehow either ceased to exist to ceased to be competitive, what do you think would happen to AMD.
There's an old adage that competition improves the breed, we should be hoping that Intel's E-core only Xeons are beasts that take the performance crown, so that AMD will be forced to improve their products again.

2) Michael, why not include that beastly Xeon Max based server you have, with all that onboard ram I'm sure Intel would have taken a few more wins.

I'm guessing maybe you are precluded from doing so by NDA.

Xeon Max isn't designed to compete with Bergamo, that's what Sierra Forest (eventually) is for. In all these highly-threaded tests the Xeon Max 9480 only falls behind the 8490H due to the lower core count, so really not of much value including it in a density/core-focused review.

[Times Two]

Two things:

...

2) Michael, why not include that beastly Xeon Max based server you have, with all that onboard ram I'm sure Intel would have taken a few more wins.

I'm guessing maybe you are precluded from doing so by NDA.

Please read the simulpublished Genoa-X review for Xeon Max results, here: https://www.phoronix.com/review/amd-...84x-benchmarks.

[NeoMorpheus]

Two things:

1) To those that think it would be great of Intel somehow either ceased to exist to ceased to be competitive, what do you think would happen to AMD.
There's an old adage that competition improves the breed, we should be hoping that Intel's E-core only Xeons are beasts that take the performance crown, so that AMD will be forced to improve their products again.

I already know what Intel will do to me and the market when they are on top (4 core decade in hell, insane pricing, illegal practices, etc), so I will be beyond dumb in giving them another chance.

I do hope they die and that should give others a chance to shine,

We have ARM, RISC-V, POWER all as possible substitute if AMD goes rogue, like Intel did.

I have zero problem if Intel, Nvidia and MS (weel, at least Windows) all died tomorrow.

[CochainComplex]

sophisticles your are totally right about 1) we can observe what happens if one competitor has almost no competition. Nvidia. So one can savely say that a healthy market needs at least 3 equally strong competitors.

[willmore]

I feel that Michael should have included some 16 core desktop AMD chips so that Intel would have had someone they could have beaten half of the time.