• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Intel Arc series unveiled with the Alchemist dGPU to arrive in Q1 2022

In a very short summary: PC World youtube video seems to put the A370M in the middle ball park of the laptop 3050 range ... but behind the 3060 series. And in the single video encoding test they were able to do, it was closer to the 3060 than the 3050, likely because it could leverage the intel CPU encoders as well (a bit like quicksync?)

Whether that is good or not will come down to price versus the 3050 / 3060's etc.

They didn't slate it, but with such a small range of tests, its hard to say much more than that without a price reference.

What cpu did they use? Iirc Intel did mention that it's GPUs are able to speed up some processing if you have an Intel CPU which has an iGPU and I think I encoding was one of those so when you encode the system uses the iGPU on the CPU to help do the encoding so that's possibly why the a370m has RTx3060 encoding performance it's because the Intel iGPU on the CPU is helping

Reviewers will need to be careful with this when benchmarking, since Intel GPUs may run faster with an Intel CPU than an AMD CPU. I wonder if Intel GPUs will be able to make use of the iGPU in Ryzen 7000 but I won't hold my breath - we're probably heading towards a return of the days where you need to pair an AMD GPU with an AMD CPU and Nvidia/Intel GPU with an Intel CPU
 
Last edited:
Rumours keep saying the delays are for driver problems lol


Who here will be brave enough to buy an Intel GPU at launch and roll the driver god dice?


The one thing we could take from this review is using the numbers to scale up the other models, the top end Arc mobile GPU has 4x number of cores compared to this one and runs with 100mhz higher base clock. If there is a linear scale for Intel the top cards fps would be 4x the numbers seen in this review

The a370m gets 4400 points in Timespy, so linear scale gives the top card the A770m 17600 points in a timespy. This is the same score that the rtx3080ti mobile GPU gets
Have a read of this post from a year ago: https://forums.overclockers.co.uk/threads/intel-xe-gpu-speculation.18929566/post-34831643
Basically what I'm saying is, yes drivers will be an issue, and people trying to 'interpolate' from early numbers will be massively disappointed. You won't be getting RTX3080tim performance, mark my words.
 
What cpu did they use? ....

Reviewers will need to be careful with this when benchmarking, since Intel GPUs may run faster with an Intel CPU than an AMD CPU. I wonder if Intel GPUs will be able to make use of the iGPU in Ryzen 7000 but I won't hold my breath - we're probably heading towards a return of the days where you need to pair an AMD GPU with an AMD CPU and Nvidia/Intel GPU with an Intel CPU

Described as a U-class, quad core, with discrete (A370M) graphics.

Regarding benchmarking ... i dont have an issue with leveraging the CPU elements to improve performance. There are always going to be benefits at times to pairing certain types of component together.
 
My thinking is also they will scrap this gen for Desktop, its too little too late

The A770M is 406mm on TSMC 6nm

For perspective the 3070 is 396mm on Samsung 8nm, the 6700XT is 335mm on TSMC 7nm.

TSMC 6nm has 18% higher density than TSMC 7nm, so the 6700XT would be 275mm on TSMC 6nm, that's a big difference in size on the same node,
RDNA3 is probably going to be clocked 15% higher and if AMD charge $399 for a 7600XT at 6700XT performance +15% Intel could only sell their 406mm GPU for $319, at that i think Intel may even lose money on everyone they sell.

This is not a good architecture, its way over sized for the level of performance, it reminds me of the Fury X, the abomination Raja Koduri dumped on AMD.
 
There is another potential problem.

AIB's use 3 different cooler designs, a large one for GA102, a mid size for GA104 and a small one for GA106.

They use the same coolers on the AMD cards, this is to keep costs down, all they do is move the mounting holes.

GA102 - Navi 21
GA104 - Navi 22
GA106 - Navi 23

Because Intel's die is so much larger than certainly the next generation AMD and Nvidia cards AIB's may be forced to use the mid size coolers for Intel GPU's that are supposed to retail in the smaller range, because the cold plates on the small tier cooler are not large enough, depending on the power consumption the cooler may not be physically enough to cool it.

So using a GH104 / Navi 32 cooler for a GH106 / Navi 33 level card, and that's an extra cost for them because those coolers are more expensive.
 
My thinking is also they will scrap this gen for Desktop, its too little too late

The A770M is 406mm on TSMC 6nm

For perspective the 3070 is 396mm on Samsung 8nm, the 6700XT is 335mm on TSMC 7nm.

TSMC 6nm has 18% higher density than TSMC 7nm, so the 6700XT would be 275mm on TSMC 6nm, that's a big difference in size on the same node,
RDNA3 is probably going to be clocked 15% higher and if AMD charge $399 for a 7600XT at 6700XT performance +15% Intel could only sell their 406mm GPU for $319, at that i think Intel may even lose money on everyone they sell.

This is not a good architecture, its way over sized for the level of performance, it reminds me of the Fury X, the abomination Raja Koduri dumped on AMD.


Raja fleeced amd now he's fleecing Intel haha the dude is a pro scammer
 
Raja fleeced amd now he's fleecing Intel haha the dude is a pro scammer

I honestly don't know why Intel hired him, he was in conflict with AMD because he wanted to syphon off AMD's GPU section for himself which caused a lot of toxic politicking with in the company, AMD relented and gave him the "Radeon Technologies Group" on top of that what he actually delivered was the Fury-X, which i think we can all agree was a bit crap.

I think Lisa gave him an ultimatum, resign or you're sacked, for as nice as she seems and sure she is she is a proper CEO, AMD had a lot of indecisive, directionless nay sayers in the CPU division and she ruthlessly purged the lot.
 
Its not uncommon for Intel to release the mobile GPU first and have it out there in the wild. Most updates I have read point to a big delay in the desktop part theorising its pushed back to Q4 but then AMD and Nvidia will probably have shown their next gen hand so to speak. Although reading about that and Nvidia selling 3xxx series alongside 4xxx points again to a lack of supply of the new gen.
 
Its not uncommon for Intel to release the mobile GPU first and have it out there in the wild. Most updates I have read point to a big delay in the desktop part theorising its pushed back to Q4 but then AMD and Nvidia will probably have shown their next gen hand so to speak. Although reading about that and Nvidia selling 3xxx series alongside 4xxx points again to a lack of supply of the new gen.

I like the idea of selling two generations of GPU at the same time if they knock the price down, i would be happy to buy a 3080 12GB at the right price while the new GPU's are out.

Its the older generation so its cheaper, sure, i'll take it.
 
I like the idea of selling two generations of GPU at the same time if they knock the price down, i would be happy to buy a 3080 12GB at the right price while the new GPU's are out.

Its the older generation so its cheaper, sure, i'll take it.

Had me wondering how that would work also. So a 3080 in stock must be cheaper than a 4080 on pre-order but then again this is one crazy world at the moment.
 
Had me wondering how that would work also. So a 3080 in stock must be cheaper than a 4080 on pre-order but then again this is one crazy world at the moment.

While the prices are still inflated, that makes sense. They can easily justify £1000 for a 4080, when a 3080 is still selling for £850, when it should be £650. Also means that it would be 4080 and higher for a while as no point in having 4060s if the 30 series is still about
 
Raja fleeced amd now he's fleecing Intel haha the dude is a pro scammer
If Raja truly is responsible for all these failures then Lisa Su did well to get rid of him along with Chris Hook and the cringy, wildly over optimistic 'poor Volta' marketing team that obviously went to Intel too (although Chris Hook was let go pretty quickly and is no longer at Intel).
 
Last edited:
I honestly don't know why Intel hired him, he was in conflict with AMD because he wanted to syphon off AMD's GPU section for himself which caused a lot of toxic politicking with in the company, AMD relented and gave him the "Radeon Technologies Group" on top of that what he actually delivered was the Fury-X, which i think we can all agree was a bit crap.

I think Lisa gave him an ultimatum, resign or you're sacked, for as nice as she seems and sure she is she is a proper CEO, AMD had a lot of indecisive, directionless nay sayers in the CPU division and she ruthlessly purged the lot.
Agreed, AMD is doing much better since Lisa made those decisions, in both graphics and CPUs. Performance; architectural innovation, progress and engineering prowess; as well as clockspeeds of GPUs greatly improved once David Wang (also formerly of ATI) took over RTG and some top CPU talent (including I think Suzanne Plummer who I rate very highly and was key in creating Zen) were moved to work with him (even though graphics still have some work to do, but I can now see them succeeding handsomely, particularly if the rumours on RDNA 3 are true).

I suspected that many of the failures were at least partly the responsibility of Raja and his team and that suspicion has only grown stronger with the way Arc seems to be turning into a disaster. I'm just glad the disaster wasn't allowed to continue at AMD. At least Intel have plenty of money to burn in getting it right whereas RTG would have been driven into bankruptcy.

Raja also delivered Vega which was massively overhyped too. The best thing that he came up with was Polaris, but even that ended up breaking the PCIE power spec so that was certainly not a smooth launch either.
 
Last edited:
While the prices are still inflated, that makes sense. They can easily justify £1000 for a 4080, when a 3080 is still selling for £850, when it should be £650. Also means that it would be 4080 and higher for a while as no point in having 4060s if the 30 series is still about

3060 is popular with system builders so I would expect the 4060 to be the same. We might end up having what we did before where only way to get a 4xxx series is to buy a system.
 
Well saw it coming, but the desktop GPUs are delayed:

https://community.intel.com/t5/Blog.../Gaming/Engineering-Arc-5-9-2022/post/1383055

System builders and OEMs in China will get A3 first in Q2, then to Chinese consumers shortly after. A5 and A7 will be later in the summer (Q3?) for worldwide system builders and OEMs, and then available to everyone after that. So... I'm expecting we probably won't see them until Q4... right against the RTX 4000 series and RX 7000 series. They better be priced aggressively if it's going to enter the market at the same time as next gen cards.
 
Have been holding out to initially get one of these as i dont need my g-card to be amazing and was hoping for a cheaper price per performance as it's their first gen, but with this further delay, it feels like getting a mid range amd/nvidia just makes more sense now.
 
Its not uncommon for Intel to release the mobile GPU first and have it out there in the wild. Most updates I have read point to a big delay in the desktop part theorising its pushed back to Q4 but then AMD and Nvidia will probably have shown their next gen hand so to speak. Although reading about that and Nvidia selling 3xxx series alongside 4xxx points again to a lack of supply of the new gen.

If nvidia was to sell both at the same time it would have to sell the 3000 series at a much lower price than the 4000 series (especially if the 4000 series is significantly more powerful, as it's expected to be). Which would mean either selling the 3000 series at the same price it is now and pricing the 4000 series much higher or selling the 3000 series at much lower prices than it is now. I don't think either would work well for nvidia because the 3000 series cards would still be good enough to result in many people choosing them over a much more expensive 4000 series card. Maybe they'll rebrand 3000 series cards as lower model number 4000 series cards, maybe with a minor change or two so they can pretend it's not just a rebrand.

Well saw it coming, but the desktop GPUs are delayed:

https://community.intel.com/t5/Blog.../Gaming/Engineering-Arc-5-9-2022/post/1383055

System builders and OEMs in China will get A3 first in Q2, then to Chinese consumers shortly after. A5 and A7 will be later in the summer (Q3?) for worldwide system builders and OEMs, and then available to everyone after that. So... I'm expecting we probably won't see them until Q4... right against the RTX 4000 series and RX 7000 series. They better be priced aggressively if it's going to enter the market at the same time as next gen cards.

Intel's dropped the ball. The question is whether they'll write the whole idea off and eat the losses or try to stay in the game as an also-ran and try to come up with a competitive product in their second generation dGPUs. This time round, Intel's top end model is going to be going up against the bottom of the midrange from AMD and nvidia, at best. Add that to the fact that AMD and nvidia provide drivers that work well and Intel provide drivers that work so badly the product couldn't even be released and any general release dGPU from Intel in this generation would have to be very cheap. Cheaper than a 4050 or 7500, certainly. My guess is that we'll only really see Intel Arc GPUs in budget prebuilts.
 
The question is whether they'll write the whole idea off and eat the losses or try to stay in the game as an also-ran and try to come up with a competitive product in their second generation dGPUs.
I can see them trying it for 2/3 gens, getting nowhere then knocking it on the head. It's a tough *tough* market to be in.
 
Maybe Alchemist will be mainly a mobile win/product for intel as i am sure low performance is not as much of big a deal, then battlemage can come in on the desktop market as I am sure i read that it was a far more impressive looking product
 
Back
Top Bottom