Intel Outperforms AMD & NVIDIA

128KBGuest
First Intel server GPU data, outperforms AMD and NVIDIA!

Intel Architecture Day 2021 is giving a lot to talk about, as the company has unveiled a lot of details about its new graphics architecture that will bring the first gaming GPUs, Alchemist, finally to the market. However, Intel has surprised in the server industry, and while these are not end-user goods, they represent a preview of what is to come in the consumer market.

In this regard, Intel has finally given figures for its Ponte Vecchio GPUs, with performance data that leave NVIDIA’s accelerator cards, which currently dominate the industry, in the lurch.

It should be noted that what Intel has released thus far of its future “Ponte Vecchio” server GPUs is preliminary data, and the device exhibited is still a prototype. Overall, and as we will show you immediately, the silicon giant’s performance stats are, to put it mildly, overwhelming. Let’s take a closer look.

Up to 45 TFLOPS in Intel Ponte Vecchio server GPUs

Intel claims that their new GPU (based on A0 silicon as they indicate) in its current state is capable of generating an output of over 45 TFLOPS in FP32 computations. To put this in context, NVIDIA’s A100 “Ampere” accelerator cards achieve “only” 19.5 TFLOPS under the same conditions, meaning that Intel has accomplished more than double the the performance of NVIDIA, and we repeat, we are only talking about preliminary data at the moment because the A0 model is still in development.

It should also be noted that with these Ponte Vecchio GPUs, Intel is not only leaving NVIDIA (very) far behind, but also AMD: its Instinct MI100 CPUs only deliver 23.1 TFLOPS of FP32 performance.

It should be noted, on the other hand, that Intel’s prototype of this server GPU was running at just 1.37 GHz, and moreover on a machine with just one of these accelerator cards and accompanied by an Intel Xeon “Sapphire Rapids” processor, also in the preliminary stage.

This is only a prototype. The commercial model will be better!

The A0 numbered silicon is always the first batch of chips to leave the foundry where they are manufactured, and for this reason we are treating it as a simple prototype. Normally, manufacturers use the first batches of chips produced to perform preliminary tests internally, and although they may sometimes send these samples to partners so that they can also perform their own, it is clear that these are preliminary models and therefore so are the performance figures they yield.

It is extremely normal for these processors to have a substantially lower running speed than the commercial models, so while the performance numbers Intel has provided so far are quite remarkable, they will eventually be lower.

Every Intel Ponte Vecchio (OAM) server GPU package generates 32,768 FP32 operations per clock cycle, Intel said during the presentation. Also, each of the two stacks is equivalent to 128 Xe cores. If each GPU vector engine does 256 operations every clock cycle, this comes up to exactly 32,768 FP32 operations for each stack of two. We can calculate that 45,000 GLOPS divided by 32,768 FP32 operations per clock cycle equals a speed of 1,373 MHz, or 1.37 GHz GPU speed, using these statistics.

With all of this, we want to point out that the data that Intel is currently handling is most likely far below what we will see once the product is ready for the consumer market, so if it is currently trailing both NVIDIA and AMD, when it hits the market in a few months, these figures could be even more shocking.And, let’s remember, enterprise/server products from companies like Intel eventually end up being reflected (at least their architecture and technology) in the consumer market as well.

GET SOCIAL!

New Console Coming? Half Life 2 on Switch & More
New Console Coming? Half Life 2 on Switch & More

=============================
Gear We Use In Our Videos
=============================
Podcast Camera: https://geni.us/XVRbQt
Podcast Lens: https://geni.us/nu1t
Podcast Mics: https://geni.us/tAdjb0
Mic Stands: https://geni.us/2hE4C
Desk Of Jingle Glory: https://geni.us/XH1R0A
Streaming Mics: https://geni.us/Uan6tfc
Streaming Camera: https://geni.us/d01j
🛒 Our Kit Profile: https://kit.co/ingaf
=============================
Music
=============================
Artlist Get 2 Months Free: https://bit.ly/2RVQGig
Twitch Music Library: https://120fps.co.uk/product/copyright-free-music-library/

=============================
Follow Us
=============================
Twitter: https://www.twitter.com/128kb_podcast
Facebook: https://www.facebook.com/128kb

=============================
Affiliate Links Notice
=============================
Some of these links are affiliate links, which means at no extra cost to you, we will make a small commission if you click them and make a purchase.

🚩 If you’re a YouTuber, you have to check out TubeBuddy. It’s an amazing tool!
https://www.tubebuddy.com/ingaf

Total
1
Shares
Previous Post

The Next Gen of RAM?

Next Post

Call Of Duty Vanguard First Impressions and Reactions

Related Posts