Intel flagship Arc Alchemist GPU has lackluster first showing in leaked benchmarks

[mongeese]

Forward-looking: Intel’s upcoming Arc flagship has made its first appearance in an online benchmark database. An unnamed Intel partner took a GPU engineering sample for a whirl in the Geekbench OpenCL test suite using a slightly dated i5-9600K validation platform and here's what we learned...

In outright numbers, the GPU does poorly. The four runs achieved an average OpenCL score of about 67,000 points and its highest score was 69,000 points. For comparison, the AMD Radeon RX 6800 XT releases about a year ago sits in the 150,000 to 180,000 point ballpark.

But don’t be discouraged. Consider this: according to the metadata collected by Geekbench during the runs, the GPU was hovering at around 1.3 GHz. At that frequency, with 512 EUs, the GPU has a theoretical 11 TFLOPS of performance.

AMD’s Radeon RX 6600 XT has similar 11 TFLOPS and scores in the 60,000 to 70,000 point range. Hence, Intel’s architecture has the same per-TFLOPS performance in OpenCL as AMD’s does.

So the question is: can Intel match AMD’s high-end TFLOPS numbers?

512 is the magic number that says this is the flagship, but it’s 512 EUs, not CUs.

According to Geekbench, the GPU had a rated clock speed of 2.1 GHz. If it had run at that speed, and not at 1.3 GHz, then it would’ve had 17 TFLOPS, which is on par with the RX 6800 (non-XT). So the answer to that question is, yes, if 2.1 GHz is achievable.

And frankly, 2.1 GHz isn’t that high. If the rumors are true and Intel is using the TSMC N6 node, then they could potentially push the clocks high enough to break into RX 6800 XT and RTX 3080 performance territory.

Remember that this is pure conjecture, but we're not the first to claim that Intel’s GPUs will land somewhere alongside the RTX 3070 or RTX 3080. Intel themselves hinted that their flagship will be priced a touch above the latter at $740 to $790. Still, as today’s OpenCL scores show, Intel may still be hard at work to get there.

Permalink to story.

https://www.techspot.com/news/92319-intel-flagship-arc-alchemist-gpu-has-lackluster-first.html

 

[theruck]

Is it april fools day already?

 

[Shadowboxer]

Whilst top end GPUs would be nice, it wouldn’t be smart. AMD recently dabbled in selling expensive high performance GPUs and they haven’t had much luck, their market share has actually gone down and there is clearly an oversupply of 6900XTs in some areas. And the reason for this is because consumers spending big money expect more features like RT, DLSS, CUDA etc. Whilst I’m sure Intel are readying features for their cards, it will take a while for these features to mature and be adopted. How many games uses XeSS for example?

The smarter decision is go after the market that AMD previously dominated but has since abandoned - the midrange value area. If this GPU has similar performance to a 6600XT but it costs $300 then it’s a massive win for Intel and consumers. And Intel can get a foothold in the market.

 

[fps4ever]

I'll be impressed if the first consumer silicon can do 3060 Ti performance out the gate with first revision drivers.

 

[Nobina]

It would be very surprising if Intels first gen GPU manages to compete with high end AMD and Nvidia cards. I doubt that's gonna happen and I doubt it's gonna bring the prices down. Only if there's a sudden influx of GPUs can that happen.

 

[Dimitriid]

Good news for miners. And yes I can say that without stuttering simply because I *still* have no faith in intel driver & feature support like, at all: If you think the Alder Lake launch with so many broken games was bad you just wait for these cards to basically be unusable for 50 to 60% of all games.

So while it might seem unfair we have been dealing with intel sup-par driver support for decades now and if the best they can do was hire the same AMD Radeon guy that kept promising better driver support and still to this day, never delivered on that promise well, you'd be a fool to seriously consider these gpus at launch. Or even several months into their launch really.

 

[paul1122]

I think they should call it the Real3D StarFighter.

 

[pcnthuziast]

Availability is the only metric that matters right now and these won't be any different than any other gpu right now. In the hands of miners and scalpers FIRST and before 'mainstream' consumers. Period. Typical consumers will need to pay hugely inflated prices regardless.

 

[Shadowboxer]

Good news for miners. And yes I can say that without stuttering simply because I *still* have no faith in intel driver & feature support like, at all: If you think the Alder Lake launch with so many broken games was bad you just wait for these cards to basically be unusable for 50 to 60% of all games.

So while it might seem unfair we have been dealing with intel sup-par driver support for decades now and if the best they can do was hire the same AMD Radeon guy that kept promising better driver support and still to this day, never delivered on that promise well, you'd be a fool to seriously consider these gpus at launch. Or even several months into their launch really.

I must say it doesn’t bode well for Intels drivers. But I have only experienced that for their iGPUs. Their discrete solutions might have better support, AMD certainly provide more support for discrete GPUs than their APUs. We will have to look on the forums to find out though because reviewers practically never comment on drivers. Reviewers seem more interested in not upsetting fanboys than showing any big differences between companies solutions. The AMD vs Nvidia driver solution is remarkably lopsided, Nvidias drivers are dramatically superior, with more features, more frequently updated, they are WHQL more often than not and play a lot nicer with Windows than Radeon drivers do but the hardware reviews rarely mention it.

 

[Lionvibez]

Ahh Geekbench the gift that keeps on giving.

There is more to GPU performance than just TFLOPS numbers.

 

[Vulcanproject]

Drivers drivers drivers. Critical aspect of modern GPU performance. Intel have to make sure their drivers are up to snuff and constantly improved otherwise world class hardware is meaningless.

 

[Dimitriid]

Availability is the only metric that matters right now and these won't be any different than any other gpu right now. In the hands of miners and scalpers FIRST and before 'mainstream' consumers. Period. Typical consumers will need to pay hugely inflated prices regardless.

If you add to this what I alluded to: there's a really big chance driver support out of the gate will be atrocious anyway, this has the potential to basically be a miner only product except for the few review samples that will basically tell customers "Even if you could buy these, you probably wouldn't want to"

 

[brucek]

The "top performer" crown is fun but what is most important here is if Intel can provide an abundant supply of well-priced GPUs that can run current games at good settings/frames.

If they can do that they'll be bringing a lot of benefit even if their halo product doesn't match Nvidia's or AMD's halo product.

 

[hahahanoobs]

AMD is praying so hard right now this is another dud. If Intel can consistently it out $200 to $600 GPUs, AMD will have no choice but to step up. It won't be easy with all their silicon coming from TSMC.

The next two years will show is what AMD (CPU and GPU) and Intel (dGPU) are really made of.

 

[Mr Majestyk]

Just in case people haven't got the message yet, the success of ARC series will be more dependent on support via frequent driver updates, than on RAW performance. I'm sure the Alchemist will be 3070 competitive in many games, but most people would rather a bit slower performance with good driver updates and rock solid experience - no BSOD issues - than great performance in a a few games and lousy drivers.

Just having a strong third player in the GPU market is the breath of fresh air we've needed for years. Now we need the same to happen in the CPU market. Let's hope Qualcomm can actually deliver on their hype about 2023 CPU's being competitive with M1. But we will need M$ to take ARM seriously.

 

[Puiu]

The biggest problem is that it's not a 2021 GPU anymore. 2022 will be big for both AMD and NVIDIA.

 

[Mr Majestyk]

The biggest problem is that it's not a 2021 GPU anymore. 2022 will be big for both AMD and NVIDIA.

3070 level performance will still be just as good in 2022 as it is in 2021. Nothing changes. I don't need 300fps when 200fps is more than enough. Couple this with the improvements in AI upscaling and these cards will have a long shelf life.

 

[Vanderlinde]

Drivers drivers drivers. Critical aspect of modern GPU performance. Intel have to make sure their drivers are up to snuff and constantly improved otherwise world class hardware is meaningless.

Let the GPU prove itself any first. It was said Vega was a power house too but more for compute then gaming relate tasks. I hope raja is not in the mix of firing out another Vega but under Intel Wings here.

And even if the GPU does house some power, I hope that could be adressed as well with drivers because intel and their drivers are'nt as rapid as Team red or green is here.

At least we have 4 different GPU vendors now consumers can start choosing of.

 

[Puiu]

3070 level performance will still be just as good in 2022 as it is in 2021. Nothing changes. I don't need 300fps when 200fps is more than enough. Couple this with the improvements in AI upscaling and these cards will have a long shelf life.

I highly doubt that it will be that good. I'm glad that the mid to low end are getting more options, but launching late is means that it won't be the 3000 series they will be competing with. Intel needed to hit it out of the park with compelling offerings this year, especially since stocks are low for AMD and Nvidia.

 

[Watzupken]

AMD is praying so hard right now this is another dud. If Intel can consistently it out $200 to $600 GPUs, AMD will have no choice but to step up. It won't be easy with all their silicon coming from TSMC.

The next two years will show is what AMD (CPU and GPU) and Intel (dGPU) are really made of.

You know that Intel GPUs are produced by TSMC right? If the GPU market was as easy as pie to step in, we won't have to wait till now for another big player to join the party. If anything, I don't think Intel can pull it off with the first few gen of their ARC GPUs. The saving grace for them is that the GPU shortage will linger for awhile, which also means if these GPUs are decent in performance and not crazy in price, they will sell out. Having said that, Intel is releasing their first gen gaming GPUs in 2022, while Nvidia and AMD are expected to announce their next gen GPUs in the same year, so I am not expecting big threat coming from Intel. If they cannot keep up with the upgrade cycle as the duopoly, chances of failure is very high over time.

 

[hahahanoobs]

You know that Intel GPUs are produced by TSMC right? If the GPU market was as easy as pie to step in, we won't have to wait till now for another big player to join the party. If anything, I don't think Intel can pull it off with the first few gen of their ARC GPUs. The saving grace for them is that the GPU shortage will linger for awhile, which also means if these GPUs are decent in performance and not crazy in price, they will sell out. Having said that, Intel is releasing their first gen gaming GPUs in 2022, while Nvidia and AMD are expected to announce their next gen GPUs in the same year, so I am not expecting big threat coming from Intel. If they cannot keep up with the upgrade cycle as the duopoly, chances of failure is very high over time.

TSMC for now to help with capacity... Already secured most of 3nm at TSMC for CPU and GPU.
Intel dGPU's will eventually move to Intel fab.
All AMD has is TSMC and GloFo for 12nm I/O dies. NVIDIA could use TSMC and Samsung again. Intel has TSMC and Intel fabs.
Guess who had more supply at launch and throughout? NVIDIA and Intel (CPU's). AMD struggled with current gen CPU and GPU stock out of the gate.

AMD will be the one to keep an eye on if ARC is good.

 

[Puiu]

TSMC for now to help with capacity... Already secured most of 3nm at TSMC for CPU and GPU.
Intel dGPU's will eventually move to Intel fab.
All AMD has is TSMC and GloFo for 12nm I/O dies. NVIDIA could use TSMC and Samsung again. Intel has TSMC and Intel fabs.
Guess who had more supply at launch and throughout? NVIDIA and Intel (CPU's). AMD struggled with current gen CPU and GPU stock out of the gate.

AMD will be the one to keep an eye on if ARC is good.

3nm won't be relevant in 2022 for AMD and, as far as I understood, Intel secured the 2022 production, which is just 10% of the yearly production that TSCM has planned for full mass production.

 

[Aceseven]

Whats the point of this article? its not even released yet and its being judged based on guesses

 

[EdmondRC]

My gut tells me that Intel's GPUs will launch underwhelming and inconsistent in benchmarks some of which will be due to driver bugs. There is no reason to believe there will not be hiccups and speed bumps for Intel with its third attempt at discrete gaming GPUs. That being said, if they can target that mid-range market and beat the MSRP of the competitors they will establish a consumer base. I look to the second generation (or third if you count DG1) to really make or break this attempt for Intel.

 

[McMurdeR]

It won't matter much if these cards are rough around the edges at launch. Intel has the capitol and engineers to refine the product line over time. Their very presence will be a big moment in gpu's, and will mark the first time in many years that the number of companies designing and selling Graphics Processors for PC gaming has increased.
Success will not necessarily be about beating Nvidia and AMD in performance first time, rather the establishment of a pipeline of design, manufacture and distribution that will enable them to compete in the coming years.