Could async compute and higher GPU utilisation cause system failures? *spawn

Rodéric

a.k.a. Ingenu
Moderator
Veteran
Asynchronous compute is about using unavoidable GPU down-time. No matter how complex the graphics, there are spells where the rendering has a lull and the GPU is twiddling its thumbs. Async compute fills those holes with work. There's definitely a good couple of hundred GFLOPS available there for a lot of titles, although that may be purposed towards compute-based graphics as much as fancy AI or physics.

Correct, but I'd be a little stressed about thermals then, have the consoles been made so that the GPU&CPU run close to 100% a fair amount of time ?
Last gen proved they don't always make the best decisions when it comes to cooling...
 
Correct, but I'd be a little stressed about thermals then, have the consoles been made so that the GPU&CPU run close to 100% a fair amount of time ?
Last gen proved they don't always make the best decisions when it comes to cooling...

I don't see why you say that. Last gen they were hit by lead-free solder transition problems. The cooling was fine.
 
I don't see why you say that. Last gen they were hit by lead-free solder transition problems.
Yep. Later models were robust. I assume the engineers have the data on thermals for running the GPUs at near 100% and designed that in to their cooling system (although mistakes do happen!).
 
Yep. Later models were robust. I assume the engineers have the data on thermals for running the GPUs at near 100% and designed that in to their cooling system (although mistakes do happen!).

As a designer you should always design for the worst case heat output, i.e. GPU running flat out, and then add in some safety margin on top for your cooling solution.

For these current-gen boxes, you may find that at 100% GPU/CPU load, the fan noise becomes a little bit more audible, but that's about it. I wouldn't expect the system to overheat, i.e. failure of the cooling system design, unless you are running the console next to a radiator on full whack, in summer, in an equatorial country, without air conditioning. But then I think you'd overheat and turn off the radiator and turn on the AC, before a sensibly designed console would.
 
Yep. Later models were robust. I assume the engineers have the data on thermals for running the GPUs at near 100% and designed that in to their cooling system (although mistakes do happen!).

you are wrong shifty. PS3 slims still use lead free solder and my 250gb slim YLOD'd after 3yrs of use. even when I changed the thermal paste , it didn't solve the ylod. BTW before this my 40gb phat ps3 ylod'd after only 1.5yrs of use. Maybe my luck with ps3s is bad.
 
Robust doesn't mean infallible. YLOD on a PS3 represents a wide range of problems and isn't always an overheating issue. The point is the catastrophic failures of early XB360s and PS3s due to the new lead-free solder didn't continue through the entire lifespan of the product. We don't still have 30% or whatever crazy number failures on new machines.
 
you are wrong shifty. PS3 slims still use lead free solder and my 250gb slim YLOD'd after 3yrs of use. even when I changed the thermal paste , it didn't solve the ylod. BTW before this my 40gb phat ps3 ylod'd after only 1.5yrs of use. Maybe my luck with ps3s is bad.
Maybe. I've never had a PS3 fail on me. I sold my launch 20GB unit after 3+ years (or whenever the first slim model was launched). And I still have the first slim model that I bought at its launch (admittedly, it hasn't been used much since I got my PS4). In fact, the only person I know personally that has had issues with a PS3 was not YLOD related but an HDD failure. Mind you, most people I know are casual gamers and I'm the only one who bought one at launch.

Anyway, nothing is fail proof... all electronics die and there will be some lemons. But it seems like the new slim models have better long-term reliability.

/OT
 
Last edited by a moderator:
Maybe. I've never had a PS3 fail on me. I sold my launch 20GB unit after 3+ years (or whenever the first slim model was launched). And I still have the first slim model that I bought at its launch (admittedly, it hasn't been used much since I got my PS4). In fact, the only person I know personally that has had issues with a PS3 was not YLOD related but an HDD failure. Mind you, most people I know are casual gamers and I'm the only one who bought one at launch.

Anyway, nothing is fail proof... all electronics die and there will be some lemons. But it seems like the new slim models have better long-term reliability.

/OT

I had a PS3 that went YSOD (young son of destruction). He put a fisher price "My first CD player" disc into my PS3, which required opening the unit to get out. Its still worked until he tried out the multi disc load capabilities of the console.
 
I had a PS3 that went YSOD (young son of destruction). He put a fisher price "My first CD player" disc into my PS3, which required opening the unit to get out. Its still worked until he tried out the multi disc load capabilities of the console.

I had a ps3 that just slowly died. The drive slowly stoped working then it would power off randomly and then started to sound like a jet plane taking off. It was a 40 gig unit
 
Will PC's be more susceptible to this kind of overheating or is that a solved problem now thanks to variable clock speeds?
 
This is an interesting question. Furmark has certainly caused thermal problems on past PC GPUs, but modern GPUs throttle automatically if the thermal budget is exceeded. Asynchronous compute can be used to utilize the GPU better. I expect to see at least some games utilizing asynchronous compute to reach Furmark-level thermals.But I don't expect to see any thermal failures or instability, because the thermal management systems of the modern GPUs have no instability problems in Furmark either.
 
Will consoles potentially throttle? I presume not and the fan will just spin up - the quietness of these machines means a lot of capacity must exist in the cooling systems as latent blowability.
 
presently looking at the builds of both ps4 and Xbox one , which one is better built of managing thermals ?
You can't tell from looking, you'd need to test each by recording the temperature of the APU cores and other parts of the system like the north and sound bridges.

The Xbox One has conventional 112mm fan/sink combo intended to disperse the heat air above the chips with heat escaping primarily through vents at the top and rear of the console. The PS4 has a 85mm centrifugal fan and cooling system designed to create specific airflow using pressure from air drawn in through low intakes at the sides up through the fan, over the chips then out of vents.
 
presently looking at the builds of both ps4 and Xbox one , which one is better built of managing thermals ?
The xbox one is obviously going to generate less heat than the ps4. I assume this is why MS were so reticent in putting a high end card in there, they were crapping themselves over another repeat of the RROD fiasco which cost them a $1Billion or two, but it's cost them heavily in terms of market share.
 
The reason's for XB1's hardware choices likely have very little to do with XB360's early failure rate. That was due to a new engineering concern no-one had mastered. Later consoles got on top of it, and that know-how can be applied to future products. XB1's BOM is already very higher, leading to a high entry cost. It's that BOM and cost/benefit balance that decided the hardware, and not a once-off, unexpected design issue that caught everyone unawares.
 
The cooling was fine.
Not on launch 360s from what I understand; the GPU ran really hot crammed in under the optical drive like it was and MS had to fix that with a band-aid extra heatpipe cooler... It surely contributed to the solder issues we saw.
 
Will consoles potentially throttle? I presume not and the fan will just spin up - the quietness of these machines means a lot of capacity must exist in the cooling systems as latent blowability.

Microsoft actually talked about what their solution to thermal issues was, but I can't remember exactly. I know the system will shut down if it reaches a certain threshold.

http://www.cnet.com/news/xbox-one-knows-when-its-overheating-adjusts-accordingly/

They ramp up fan first, and then start lowering clock speeds if that doesn't work.
 
Back
Top