News:

Willkommen im Notebookcheck.com Forum! Hier können sie über alle unsere Artikel und allgemein über Notebook relevante Dinge disuktieren. Viel Spass!

Main Menu

Powerful AMD Strix Point APUs with 9 TFLOP RDNA 3+ iGPU and hybrid Zen 5 CPU could make low-end discreet GPUs "irrelevant"

Started by Redaktion, January 16, 2023, 17:00:07

Previous topic - Next topic

Redaktion

AMD Strix Point APUs are slated to release in 2024. The chips reportedly sport a hybrid architecture through a combination of big and small CPU cores. Serial leaker Paul from RedGamingTech now alleges that the Strix Point's iGPU has the potential to challenge discreet lower-end graphics cards.

https://www.notebookcheck.net/Powerful-AMD-Strix-Point-APUs-with-9-TFLOP-RDNA-3-iGPU-and-hybrid-Zen-5-CPU-could-make-low-end-discreet-GPUs-irrelevant.682022.0.html

Russel

Red fanatic smoking pot and spewing his 'revelations' xD.
AMD's management can't even get their cards decent cooling solution or get their chips into premium designs. And they're taking all credit for what the engineering team has done while not doing their job.
Engineers already delivered. Zen 2 was superior to anything intel offered, but how many design wins and ACTUAL laptops have you seen with zen 2 and up?

Anonymousgg

Sounds more like a laundry list of what people want in Strix that didn't show up in Phoenix than a leak.

If it uses an 8+8 core design, and more than 16 CUs, then I'll be very impressed and use a desktop version.

Fabricio

I doubt it will be that powerful igpus have just achieved 4tf with the 660m and desktop igups are much weaker a leap that big is unbealiveable

Anonymousgg

Quote from: Fabricio on January 16, 2023, 22:21:20I doubt it will be that powerful igpus have just achieved 4tf with the 660m and desktop igups are much weaker a leap that big is unbealiveable

I think you meant 680M, which is more like 3.4 teraflops.

Phoenix's 780M could be the equivalent of almost 5 teraflops, and then it's entirely within AMD's ability to improve by 80% with Strix Point. They can't be counted on to do it, but they could choose to make APUs a lot faster as part of a strategy to kill sub-$200 graphics. Almost nobody using 1080p resolution will need discrete graphics.

AMD desktop APUs will get the same iGPU as mobile APUs, whenever they finally appear on AM5.

D

I think it's more possible than ever that low end boards could become a thing of the past.  First, low end desktops won't need them as AMD'S RDNA 3+ and later RDNA 4 equipped APUs will be all that is needed for 90% of computer users needs who aren't gamers.  And it reduces the BOM (Billable Order of Materials) for OEMs thus lowering costs and/or increasing profit margins.  This also includes the small form factor PCs (think Mac Mini sized PC's).  Because laptops sales outpace desktops of any size this also decreases the relevancy of low end discrete GPUs.  I also tend to believe the claims of desktop like iGPU performance because it's not all about the teraflops and the gigahertz.  It's also about the way AMD is using Infinity Architecture and heterogenous memory management in both Windows and Linux to make it where there is no longer the need to make duplicate copies of data in both the CPU and the GPU which leads to latency and stalls.  DDR5 along with the upcoming PCI 5 will be not only faster but more atomic.  These efficiencies and more will accentuate the speed gains in the iGPU to the point that low end boards will become more and more irrelevant.  I would also imagine that Strix becomes the basis for (highly modified of course) for the PlayStation 6 whenever that occurs.

Duds

Quote from: Anonymousgg on January 16, 2023, 23:28:25
Quote from: Fabricio on January 16, 2023, 22:21:20I doubt it will be that powerful igpus have just achieved 4tf with the 660m and desktop igups are much weaker a leap that big is unbealiveable

I think you meant 680M, which is more like 3.4 teraflops.

Phoenix's 780M could be the equivalent of almost 5 teraflops, and then it's entirely within AMD's ability to improve by 80% with Strix Point. They can't be counted on to do it, but they could choose to make APUs a lot faster as part of a strategy to kill sub-$200 graphics. Almost nobody using 1080p resolution will need discrete graphics.

AMD desktop APUs will get the same iGPU as mobile APUs, whenever they finally appear on AM5.

It is a common mistake to assume that AMD or even Intel can just make their integrated GPU's faster. They are limited by memory bandwidth not by the size of the GPU.

Switching to DDR5 gave APU's a really nice generational bump but there'll be a bit of a wait for a new memory standard.

You can add more cache or you can add HBM but both of these massively add to the cost and APU's are supposed to be the budget option.

Fabricio

Quote from: Anonymousgg on January 16, 2023, 23:28:25
Quote from: Fabricio on January 16, 2023, 22:21:20I doubt it will be that powerful igpus have just achieved 4tf with the 660m and desktop igups are much weaker a leap that big is unbealiveable

I think you meant 680M, which is more like 3.4 teraflops.

Phoenix's 780M could be the equivalent of almost 5 teraflops, and then it's entirely within AMD's ability to improve by 80% with Strix Point. They can't be counted on to do it, but they could choose to make APUs a lot faster as part of a strategy to kill sub-$200 graphics. Almost nobody using 1080p resolution will need discrete graphics.

AMD desktop APUs will get the same iGPU as mobile APUs, whenever they finally appear on AM5.

I agree theoretically they could make igpus based on the consoles ps5(9-10tf) xsx(12tf) which are both zen 2 and rdna '1.5' and 2 however it wouldnt be financially viable since it would be much more expensive and require a much more beefier power supply and cooler so getting igpus at this level only in 5 or so years
Quote from: Duds on January 17, 2023, 04:01:51
Quote from: Anonymousgg on January 16, 2023, 23:28:25
Quote from: Fabricio on January 16, 2023, 22:21:20I doubt it will be that powerful igpus have just achieved 4tf with the 660m and desktop igups are much weaker a leap that big is unbealiveable

I think you meant 680M, which is more like 3.4 teraflops.

Phoenix's 780M could be the equivalent of almost 5 teraflops, and then it's entirely within AMD's ability to improve by 80% with Strix Point. They can't be counted on to do it, but they could choose to make APUs a lot faster as part of a strategy to kill sub-$200 graphics. Almost nobody using 1080p resolution will need discrete graphics.

AMD desktop APUs will get the same iGPU as mobile APUs, whenever they finally appear on AM5.

It is a common mistake to assume that AMD or even Intel can just make their integrated GPU's faster. They are limited by memory bandwidth not by the size of the GPU.

Switching to DDR5 gave APU's a really nice generational bump but there'll be a bit of a wait for a new memory standard.

You can add more cache or you can add HBM but both of these massively add to the cost and APU's are supposed to be the budget option.

Yes thats why often amd gpus with similar teraflops(raw power) end up having a significant lead in some cases against similar igpus thats also one of many advantages consoles apus have

Erm

I don't think low-end discreet GPUs need any help in making themselves irrelevant. They're already doing a mighty fine job of that on their own. Aren't RTX 3050's for example almost $400 at some places / in many regions? Not too long ago, that used to be the price of GTX 1070.

Also, wasn't RGT the same guy saying how RDNA3's Navi 31 would be 3x as fast compared to previous gen, reach over 3 Ghz and how Nvidia were in trouble / RTX 4000 series would be crushed along with many other speculative youtubers? And we all know how that ended up. There's a bit too much fluff coming out of these guys.

At this point, I'm honestly more excited about future Arc GPU's. At least you don't need to wait 2 years to find one in stock at reasonable a price with Intel. The current Nvidia/AMD market is a joke. (Both for APU/GPU and desktop/laptop)

Might as well just get a console/macbook rather than PC for now. It just makes more sense as it's gonna take awhile for Intel to release Battlemage and further refine their driver stack.

digitalmarketingcourse

I read this post your post so nice and very informative post thanks for sharing this post. Do visit our site which provide the Best Digital Marketing Course in Raipur

RipOffEurope

50 class GPUs only exist to have a low "starting at" price and fool average consumers who buy uninformed

Anonymousgg

Quote from: Duds on January 17, 2023, 04:01:51It is a common mistake to assume that AMD or even Intel can just make their integrated GPU's faster. They are limited by memory bandwidth not by the size of the GPU.

Switching to DDR5 gave APU's a really nice generational bump but there'll be a bit of a wait for a new memory standard.

You can add more cache or you can add HBM but both of these massively add to the cost and APU's are supposed to be the budget option.

There's surely more gas left in the tank after Rembrandt/Phoenix. And maybe that will have to be assisted by faster memory. Rembrandt officially supported DDR5-4800 and LPDDR5-6400, but we've already seen talk of LPDDR5X-8500, LPDDR5X-10000, and DDR5-12600, for example. DDR5's changes to power/voltage management are going to allow massive increases in speeds over the entire generation. LPDDR5 already has its successor ready for 2023-2024 products.

I really want to see big cache and L4 on APUs, but it's a departure from what AMD wants to do. If they can design monolithic dies to have TSV connection points, maybe they can put an optional cache on only a top model, such as the equivalent of a Ryzen 9 6980HS or Ryzen 7 5700G (which was not originally a budget option with a retail MSRP of $359). Only a fully enabled and highly clocked model would need all the memory bandwidth anyway, not the models with half the CUs.

I don't believe AMD should be expected to go in this direction, but if they want to replace low-end discrete graphics on mobile/desktop with APUs, they have ways to do it.

Ty

Definitely haven't been hearing this claim since Llano launched in 2011...

If you're not playing demanding or newish games it's fine. Try to play something from the last 5 years at 1080p and even medium detail settings, the large textures and everything else the computer is doing will quickly suffocate the APU. The shared system memory bandwidth is always the limiting factor.

Travis McMurry

What did this APU do that requires discretion?  Is it doing something shady?  Does it have a mister/mistress? 

To avoid silly questions like this in the future, and be more clear about the point of the article, please use the correct word "discrete" and not discreet. 

Discrete - apart or detached from others; separate; distinct; as in "The GPU is a discrete part on the motherboard"
Discreet - judicious in one's conduct or speech, especially with regard to respecting privacy or maintaining silence about something of a delicate nature; as in "AMD should be more discreet with whom it entrusts secrets to"

D

Bandwidth limitation from DDR5. Pointless.
Even low-end GPUs have >128 GB/s. It's DOUBLE data rate RAM and DOUBLE channel, after all...
Besides, expect high prices...
Remember 5700G at launch and many months after?

Quick Reply

Warning: this topic has not been posted in for at least 120 days.
Unless you're sure you want to reply, please consider starting a new topic.

Name:
Email:
Verification:
Please leave this box empty:

Shortcuts: ALT+S post or ALT+P preview