AMD Raven Ridge mobile graphics are faster than the Intel Iris Plus 640

[Okjoek]

18 minutes ago, The Benjamins said:

ya, its all up to HBM cost. but it would be nice to see small gaming/CAD PC's the size of a NUC (or a bit bigger)

 

The biggest APU issue is GPU ram speed, and HBM makes a lot of sense.

Never bothered to ask about this before, but what is the diference from using DDR4 system RAM to something like GDDR5 or even HBM2?

[The Benjamins]

1 minute ago, Okjoek said:

Never bothered to ask about this before, but what is the diference from using DDR4 system RAM to something like GDDR5 or even HBM2?

GPU tasks need faster ram. DDR3 is about 15-25GBps DDR4 is about 20-30GBps (I might be off by a bit but close enough) were HBM1 is 500GBps 4 stacks HBM2 is 480 GBps 2 stacks (fiji, vega). GDDR5 has been seen to be anywhere in 200GBps - 500GBps

 

DDR4 will have some nice gains on the A12-9800 vs the A10-7870k (I have that in my oil PC)

 

Testing on the A10-7870k noticed 5-10% gains in FPS when going from 1600 to 1866 DDR3.

 

This is why the console APU's use GDDR5.

 

So a APU with on package HBM would eliminate the bottleneck of slow GPU RAM vs GPU using shared system RAM.

[Doobeedoo]

Expected, Raven Ridge is quite a step up in APU department, these will be awesome in laptops. Very good performance for relatevly low priced and thin laptops. 

[Guest]

1 minute ago, The Benjamins said:

So a APU with on package HBM would eliminate the bottleneck of slow GPU RAM vs GPU using shared system RAM.

It will be interesting to see if/when HBM comes to an APU, whether for consumers or businesses, if other components will gain any noticeable benefits through HSA.

[Fahim Foysal]

Finally some good news. Amd should release six core laptop cpus though.

[Guest]

5 minutes ago, Fahim Foysal said:

Finally some good news. Amd should release six core laptop cpus though.

Perhaps when they are based on zen2 cores. Those are rumored to have 6 cores per ccx.

[Taf the Ghost]

1 hour ago, tjcater said:

Perhaps when they are based on zen2 cores. Those are rumored to have 6 cores per ccx.

The APU design uses a CCX, but it's a slightly different full design. At least that's the information we have. This gives them the option to use whatever sizes they feel like with the CCX in the future. (Though I highly doubt there's much room beyond 4c/8t for a while in that space, especially on the Laptop side of things.)

[AnonymousGuy]

AMD...still trying to make people care about graphics performance in a platform that no one does any serious gaming on anyways.

 

Here's the flow chart:

 

Does integrated graphics play 4K youtube without stuttering?

 

Yes -> good enough.

[Prysin]

5 hours ago, Sauron said:

Because vega draws a lot of power. It may not matter as much on a desktop, but on a laptop it's crucial.

that is for VGA clocks. APU clocks could come down to 1100-1300MHz and still be damn fast for iGPU to be.

[Okjoek]

3 hours ago, The Benjamins said:

GPU tasks need faster ram. DDR3 is about 15-25GBps DDR4 is about 20-30GBps (I might be off by a bit but close enough) were HBM1 is 500GBps 4 stacks HBM2 is 480 GBps 2 stacks (fiji, vega). GDDR5 has been seen to be anywhere in 200GBps - 500GBps

 

DDR4 will have some nice gains on the A12-9800 vs the A10-7870k (I have that in my oil PC)

 

Testing on the A10-7870k noticed 5-10% gains in FPS when going from 1600 to 1866 DDR3.

 

This is why the console APU's use GDDR5.

 

So a APU with on package HBM would eliminate the bottleneck of slow GPU RAM vs GPU using shared system RAM.

For an APU (a product based around value) I can't see more than 2GB GDDR5 being a good choice.

 

I'm curious have you seen any real-world improvement on the AM4 Bristol Ridge APU moving from DDR3 to 4?

[The Benjamins]

10 minutes ago, Okjoek said:

For an APU (a product based around value) I can't see more than 2GB GDDR5 being a good choice.

 

I'm curious have you seen any real-world improvement on the AM4 Bristol Ridge APU moving from DDR3 to 4?

APU are not value only products, even though AMD has yet to make one that is not. AMD has a whight paper for a APU that has 4 CPU dies, 8 GPU dies and HBM2 stacks per GPU die.

 

I think the AM4 Bristol Ridge APU does get a improvement but no one cares enough to test it, sense the Raven Ridge APU's are just around the corner.

[vanished]

ok, but are they better than the Iris Pro 580?  I believe that's their best iGPU at the moment

[Okjoek]

1 minute ago, Ryan_Vickers said:

ok, but are they better than the Iris Pro 580?  I believe that's their best iGPU at the moment

I think this article was comparing against Intel's mobile offerings. Isn't Iris Pro 580 not for mobile? I think it's what's used in the Skullcanyon NUC, but IDK if that's considered "mobile" though.

[vanished]

Just now, Okjoek said:

I think this article was comparing against Intel's mobile offerings. Isn't Iris Pro 580 not for mobile? I think it's what's used in the Skullcanyon NUC, but IDK if that's considered "mobile" though.

It seems pretty mobile  not a large thing to carry around, and it's not like there's much room for cooling in those so if it works there it would probably work in a laptop.

 

But mainly I was just thinking how it stacks up out of curiosity.  Is there any other news about AMD's best iGPU?

[Okjoek]

34 minutes ago, Ryan_Vickers said:

It seems pretty mobile  not a large thing to carry around, and it's not like there's much room for cooling in those so if it works there it would probably work in a laptop.

 

But mainly I was just thinking how it stacks up out of curiosity.  Is there any other news about AMD's best iGPU?

IDK. Quite frankly I'd like to see some innovation in the laptop market. If we can have skullcanyon barebones, why can't we get a laptop shell for something like an mSTX motherboard that can take a socketed APU?

[vanished]

Just now, Okjoek said:

IDK. Quite frankly I'd like to see some innovation in the laptop market. If we can have skullcanyon barebones, why can't we get a laptop shell for something like an mSTX motherboard that can take a socketed APU?

That would be interesting.  Even just getting Ryzen laptops at all would be good imo, but I suppose they're waiting for APUs since when was the last time you saw a laptop that didn't have an iGPU for power purposes?

[Mira Yurizaki]

17 minutes ago, Ryan_Vickers said:

That would be interesting.  Even just getting Ryzen laptops at all would be good imo, but I suppose they're waiting for APUs since when was the last time you saw a laptop that didn't have an iGPU for power purposes?

Assuming they even use it. I have a suspicion MSI and ASUS don't really give a damn about using it if they put a discrete GPU in their laptops. Otherwise I can't explain why Dell and Razer make them look pathetic in the battery life department.

[vanished]

Just now, M.Yurizaki said:

Assuming they even use it. I have a suspicion MSI and ASUS don't really give a damn about using it if they put a discrete GPU in their laptop. Otherwise I can't explain why Dell and Razer make them look pathetic in the battery life department.

To be honest I'd rather just have 1 GPU and not have to worry about all the switching nonsense, but I know from experience that that will really kill battery life so I think they pretty much have to "use it"

[Mira Yurizaki]

2 minutes ago, Ryan_Vickers said:

To be honest I'd rather just have 1 GPU and not have to worry about all the switching nonsense, but I know from experience that that will really kill battery life so I think they pretty much have to "use it"

Or why can't Intel release a CPU without a GPU, especially since people use some of their SKUs with a video card 90%+ of time.

[vanished]

Just now, M.Yurizaki said:

Or why can't Intel release a CPU without a GPU, especially since people use some of their SKUs with a video card 90%+ of time.

On the desktop, I totally agree, I've been wanting this for years.  My best idea for how to decide what gets it and what doesn't is to not include the iGPU on any "k" or "x" chip.  I don't know how much cheaper Intel would be wiling to sell them if they did this, but any savings are probably worth it for those of us who literally never use it.

[KenjiUmino]

8 hours ago, System Error Message said:

excellent, cant wait for a laptop with a good AMD APU, vega dedicated GPU, 2 thunderbolt 3 slots to make use of AMD's crossfire support for using both IGP and dedicated GPU together.

 

8 hours ago, hey_yo_ said:

So if I'm a student in the market for a new XPS 13 laptop and one is Intel and one is AMD, the AMD laptop maybe slightly cheaper but the Intel one is more "future proofed" because of TB3.

 

good plan so far ... and then you realize that eGPU enclosure + graphics card combined costs more than the whole laptop did or building a gaming desktop would. 

 

been there, almost done that. ended up building a tiny ITX gaming desktop instead. didn't regret it

 

what did i learn from that: eGPU solutions are too expensive for what they can do (you are still stuck with a lowball laptop CPU)

[System Error Message]

1 hour ago, KenjiUmino said:

 

good plan so far ... and then you realize that eGPU enclosure + graphics card combined costs more than the whole laptop did or building a gaming desktop would. 

 

been there, almost done that. ended up building a tiny ITX gaming desktop instead. didn't regret it

 

what did i learn from that: eGPU solutions are too expensive for what they can do (you are still stuck with a lowball laptop CPU)

Actually the eGPU solution + laptop would be cheaper than a laptop with equivalent GPU. The problem however are companies like razer releasing a $500 dock but dell alienware AGA costs $150 ($200 from dell), and some companies are releasing thunderbolt 3 eGPU solutions for below $300.

 

However after using the alienware AGA i absolutely hate it because of the bios disabling onboard GPU. Since the bios checks for GPU and such i cant use other cards with the AGA like NICs.

[Okjoek]

1 hour ago, KenjiUmino said:

 

good plan so far ... and then you realize that eGPU enclosure + graphics card combined costs more than the whole laptop did or building a gaming desktop would. 

 

been there, almost done that. ended up building a tiny ITX gaming desktop instead. didn't regret it

 

what did i learn from that: eGPU solutions are too expensive for what they can do (you are still stuck with a lowball laptop CPU)

 

3 minutes ago, System Error Message said:

Actually the eGPU solution + laptop would be cheaper than a laptop with equivalent GPU. The problem however are companies like razer releasing a $500 dock but dell alienware AGA costs $150 ($200 from dell), and some companies are releasing thunderbolt 3 eGPU solutions for below $300.

 

However after using the alienware AGA i absolutely hate it because of the bios disabling onboard GPU. Since the bios checks for GPU and such i cant use other cards with the AGA like NICs.

I'm not interested in eGPU because I watched a vid by Jay a while back where he showed just how much the connection bottlenecks the GPU. I'd rather just get a halfway decent integrated GPU.

 

Poor people can be enthusiasts too!

 

 

[System Error Message]

1 minute ago, Okjoek said:

 

I'm not interested in eGPU because I watched a vid by Jay a while back where he showed just how much the connection bottlenecks the GPU. I'd rather just get a halfway decent integrated GPU.

 

Poor people can be enthusiasts too!

 

did you watch linus's video comparing alienware aga and thunderbolt 3? alienware AGA uses the CPU's PCIe lanes which helps to reduce or eliminate bottlenecks. Newer laptops will have thunderbolt 3 on the CPU, infact i think apple already does this. They had a macbook with 4 thunderbolt 3 slots CPU connected as there was no onboard dedicated GPU so 16 lanes to go to all 4 thunderbolt 3s.

 

Even on the alienware, it still has 4 lanes that isnt used from the CPU which could be given to its thunderbolt 3, i just dont have a thunderbolt 3 eGPU solution to compare with.

[Brooksie359]

12 hours ago, Sauron said:

I'm pretty sure Intel is significantly bigger than both of AMD's branches combined. AMD's only real advantage in that sense is their more refined drivers (since they've been in the gpu game longer), but technically there's no reason Intel couldn't make a gpu that's at least close to vega.

I would argue that the main issue is intel has to try and make a gpu without using any of AMD or Nvidia ip which is basically impossible to do while making a decent gpu. I mean I believe Intel even has licences some of AMD ip recently.