Welcome, Unregistered.

If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below.

Showing results 1 to 25 of 500
Search took 1.22 seconds.
Search: Posts Made By: Andrew Lauritzen
Forum: Console Technology 27-Aug-2014, 16:44
Replies: 632
Views: 86,780
Posted By Andrew Lauritzen
Sure I'm not saying it would be trivial to do in...

Sure I'm not saying it would be trivial to do in the consoles or anything, I'm just rejection the assertion that coherence fundamentally must be "really slow" (i.e. his last comment about the PC). I...
Forum: Console Technology 27-Aug-2014, 16:20
Replies: 632
Views: 86,780
Posted By Andrew Lauritzen
Just because coherence on the consoles is slow...

Just because coherence on the consoles is slow (i.e. external bus) does not mean it has to be that way. On an integrated CPU/GPU it really does not need to be any more expensive than multi-core...
Forum: 3D Hardware, Software & Output Devices 27-Aug-2014, 16:07
Replies: 71
Views: 11,782
Posted By Andrew Lauritzen
I'll add a major reason from my point of view......

I'll add a major reason from my point of view... because a much larger majority of games go through a small number of engines these days than in the past. That is a huge reason why it makes sense to...
Forum: 3D Architectures & Chips 27-Aug-2014, 06:32
Replies: 158
Views: 16,973
Posted By Andrew Lauritzen
It's not about the scheduling, it is literally...

It's not about the scheduling, it is literally about tracing the path through the ideal data dependency graph. The graph being wide doesn't affect that you ultimately still need something to rip...
Forum: 3D Hardware, Software & Output Devices 27-Aug-2014, 01:04
Replies: 71
Views: 11,782
Posted By Andrew Lauritzen
It should also be noted that the responses he...

It should also be noted that the responses he posted there were paraphrased so don't read too much into the exact wording.


That's not exactly true... although I admit that "safe" was probably...
Forum: Console Technology 23-Aug-2014, 22:25
Replies: 632
Views: 86,780
Posted By Andrew Lauritzen
Well... it can be measured directly with stuff...

Well... it can be measured directly with stuff like Vtune or similar of course, at least for anything x86. But ultimately the IPC of a chip isn't really as interesting as the realized performance (at...
Forum: 3D Architectures & Chips 22-Aug-2014, 01:07
Replies: 158
Views: 16,973
Posted By Andrew Lauritzen
Right but there's no compelling reason to do that...

Right but there's no compelling reason to do that in the (very near) future... honestly that amount of buffering is a symptom of the current APIs being designed for discrete, which is entirely my...
Forum: 3D Architectures & Chips 21-Aug-2014, 19:41
Replies: 158
Views: 16,973
Posted By Andrew Lauritzen
I'm not disputing that. But this thread is about...

I'm not disputing that. But this thread is about the future, and I don't think your current experience with these architectures necessarily predicts that very well frankly for the reasons that I've...
Forum: 3D Architectures & Chips 20-Aug-2014, 17:42
Replies: 158
Views: 16,973
Posted By Andrew Lauritzen
Obviously I'm not talking about practically...

Obviously I'm not talking about practically getting "infinite" parallel performance, as if such a thing were even possible. My point was that these bottlenecks exist *now* in code that runs on...
Forum: 3D Architectures & Chips 20-Aug-2014, 05:00
Replies: 158
Views: 16,973
Posted By Andrew Lauritzen
Sure, but we're talking about a case where not a...

Sure, but we're talking about a case where not a single hardware thread on a given execution unit has any work to do for many cycles. Stalled threads can't be reused for async compute, so it really...
Forum: 3D Architectures & Chips 20-Aug-2014, 00:08
Replies: 158
Views: 16,973
Posted By Andrew Lauritzen
I'm well aware of distributed scheduling, work...

I'm well aware of distributed scheduling, work stealing, etc. and have advocated it myself for many years. But there are always serial bottlenecks in there, even if it's just the atomic that you need...
Forum: 3D Architectures & Chips 19-Aug-2014, 18:23
Replies: 158
Views: 16,973
Posted By Andrew Lauritzen
Not convinced. It puts off the issue slightly on...

Not convinced. It puts off the issue slightly on some architectures, but it's hardly a holy grail as it does nothing to address the real thread of dependent work through the rendering pipeline. You...
Forum: 3D Architectures & Chips 19-Aug-2014, 16:17
Replies: 158
Views: 16,973
Posted By Andrew Lauritzen
On the one hand that looks like the most stark...

On the one hand that looks like the most stark example of big GPU/small CPU (i.e. the 15W 2 core + GT3 part) while the vast majority of parts are more skewed towards the CPU (4+2, 4+3, 2+2). I'm also...
Forum: 3D Architectures & Chips 18-Aug-2014, 23:21
Replies: 158
Views: 16,973
Posted By Andrew Lauritzen
I think when I did the math a while back consoles...

I think when I did the math a while back consoles were ~20:1 on GPU:CPU SP flops. Haswell SoCs (GT3e for the biggest GPU) are ~2:1 at best. r290 + 4790k is ~10:1. Of course HSW-E will shift that a...
Forum: 3D Hardware, Software & Output Devices 18-Aug-2014, 22:08
Replies: 55
Views: 4,183
Posted By Andrew Lauritzen
Everything has to go through WDDM to draw on the...

Everything has to go through WDDM to draw on the swap chain, etc. And the OS owns the GPU memory on Windows, so that includes OpenGL. See for instance:...
Forum: 3D Architectures & Chips 18-Aug-2014, 21:27
Replies: 158
Views: 16,973
Posted By Andrew Lauritzen
Sure, and for most GPUs it's similar. But it's a...

Sure, and for most GPUs it's similar. But it's a very "bad" core compared to a CPU as far as extracting IPC goes. And no, that's not just because "that code is bad" (although of course there's never...
Forum: 3D Hardware, Software & Output Devices 15-Aug-2014, 23:49
Replies: 71
Views: 11,782
Posted By Andrew Lauritzen
Thanks! Yeah it was fun to show off DX12 on a...

Thanks! Yeah it was fun to show off DX12 on a tablet :)


Probably doesn't surprise you that I can't really speak to that. Sorry.
Forum: Console Technology 09-Aug-2014, 21:04
Replies: 632
Views: 86,780
Posted By Andrew Lauritzen
Also true on Intel FWIW. NVIDIA is the only IHV I...

Also true on Intel FWIW. NVIDIA is the only IHV I know that actually packs the data, and I'm not even sure if they still do (probably have both options). APIs are a bit behind the times on this issue.
Forum: Processor & Chipset Technology 03-Aug-2014, 03:24
Replies: 146
Views: 20,596
Posted By Andrew Lauritzen
Skylake* guys... we've already had the two...

Skylake* guys... we've already had the two "*wells" :)
Forum: 3D Technology & Algorithms 03-Jul-2014, 01:25
Replies: 217
Views: 25,908
Posted By Andrew Lauritzen
Uhh, I'm not sure I buy that. I'm not a display...

Uhh, I'm not sure I buy that. I'm not a display pipe expert but my understanding is there are physically only N (2-3 or whatever) clock generators in most GPUs, hence the requirement to have...
Forum: 3D Technology & Algorithms 02-Jul-2014, 19:26
Replies: 217
Views: 25,908
Posted By Andrew Lauritzen
Wouldn't it be more likely related to the sync'd...

Wouldn't it be more likely related to the sync'd clock signal stuff that was part of the original restrictions for needing a DP monitor? i.e. I imagine if you had all display-port monitors it could...
Forum: 3D Technology & Algorithms 02-Jul-2014, 19:22
Replies: 3
Views: 2,000
Posted By Andrew Lauritzen
Probably is related to doing "blobby"-type...

Probably is related to doing "blobby"-type occlusion type things like your googling implies.
Forum: 3D Hardware, Software & Output Devices 30-Jun-2014, 21:58
Replies: 391
Views: 35,350
Posted By Andrew Lauritzen
PRT/tiled resources does not allocated pages on...

PRT/tiled resources does not allocated pages on the fly - all the allocating/mapping is still done from the CPU between frames, etc. Hitting real page faults or otherwise stopping GPU execution to...
Forum: 3D Hardware, Software & Output Devices 30-Jun-2014, 18:06
Replies: 391
Views: 35,350
Posted By Andrew Lauritzen
Even if the hardware supports haulting on a page...

Even if the hardware supports haulting on a page fault like that (it may or may not, I don't know), it's likely to be reeaaaallly slow. Definitely not something you want to be doing in the middle of...
Forum: 3D Hardware, Software & Output Devices 30-Jun-2014, 00:57
Replies: 391
Views: 35,350
Posted By Andrew Lauritzen
Yes, but you don't really have any good options...

Yes, but you don't really have any good options once you run out of memory. Basically in the case of "too much transparency" you just have to drop fragments. This obviously produces artefacts and...
Showing results 1 to 25 of 500

 
Forum Jump

All times are GMT +1. The time now is 12:16.


Powered by vBulletin® Version 3.8.6
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.