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 4.23 seconds.
Search: Posts Made By: OpenGL guy
Forum: 3D Architectures & Chips 16-May-2013, 15:21
Replies: 41
Views: 2,019
Posted By OpenGL guy
Execution mask, stack and condition flags are...

Execution mask, stack and condition flags are stored in registers, no need to save them when switching to another wavefront.
Forum: 3D & Semiconductor Industry 06-May-2013, 06:31
Replies: 698
Views: 87,595
Posted By OpenGL guy
Installing both drivers shouldn't affect anything.

Installing both drivers shouldn't affect anything.
Forum: 3D & Semiconductor Industry 04-May-2013, 17:01
Replies: 698
Views: 87,595
Posted By OpenGL guy
Are you using the integrated graphics for...

Are you using the integrated graphics for display? What about Virtu?
Forum: 3D & Semiconductor Industry 19-Apr-2013, 19:53
Replies: 127
Views: 5,123
Posted By OpenGL guy
Microsoft improved VidMM in Windows 7. That had...

Microsoft improved VidMM in Windows 7. That had a large impact for certain cases, although the benefits could vary depending on what GPU you had.
Forum: 3D Architectures & Chips 02-Apr-2013, 23:40
Replies: 6,520
Views: 915,583
Posted By OpenGL guy
And where would you put your DVD/Blu-ray...

And where would you put your DVD/Blu-ray reader/burner?
Forum: 3D Architectures & Chips 01-Apr-2013, 16:17
Replies: 4
Views: 1,049
Posted By OpenGL guy
That's uncoalesced accesses within a single...

That's uncoalesced accesses within a single channel. Of course separate channels can process independent addresses.
Forum: 3D Architectures & Chips 26-Mar-2013, 04:19
Replies: 1,292
Views: 179,503
Posted By OpenGL guy
My first computer was a used Apple ][. My dad...

My first computer was a used Apple ][. My dad even picked up a CP/M card (Z80-B with 128KB of RAM) for it thinking it would be compatible with his Vector Graphics machine at work (it wasn't). The...
Forum: 3D Architectures & Chips 26-Mar-2013, 02:01
Replies: 1,292
Views: 179,503
Posted By OpenGL guy
Not sure why they would have used those colors. ...

Not sure why they would have used those colors. There are other colors to choose from, see http://en.wikipedia.org/wiki/Color_Graphics_Adapter

Edit: ah I see now. You only got 16 colors in text...
Forum: 3D Architectures & Chips 26-Mar-2013, 01:07
Replies: 1,292
Views: 179,503
Posted By OpenGL guy
Magenta = full blue + full red Cyan = full blue...

Magenta = full blue + full red
Cyan = full blue + full green
Yellow = full red + full green

These are also known as the subtractive primary colors. Makes perfect sense if you want a simple...
Forum: 3D Architectures & Chips 18-Mar-2013, 16:04
Replies: 144
Views: 7,656
Posted By OpenGL guy
Yes and no. OpenCL, unlike DirectCompute,...

Yes and no. OpenCL, unlike DirectCompute, doesn't require that the application specify the workgroup size for a kernel at compilation time. To avoid recompilation, we use a default that is as large...
Forum: 3D Architectures & Chips 18-Mar-2013, 15:06
Replies: 144
Views: 7,656
Posted By OpenGL guy
Barriers are one advantage. Another is that if...

Barriers are one advantage. Another is that if the machine is full, the machine can schedule a new workgroup sooner if the workgroup is smaller as you only need to wait for a single wavefront to...
Forum: 3D Architectures & Chips 18-Mar-2013, 14:43
Replies: 144
Views: 7,656
Posted By OpenGL guy
No, that's not what I am saying. What I am...

No, that's not what I am saying. What I am saying is that it is generally more efficient to use workgroups of 64 threads.
Forum: 3D Architectures & Chips 17-Mar-2013, 15:22
Replies: 144
Views: 7,656
Posted By OpenGL guy
The 4 vector units in a CU can process separate...

The 4 vector units in a CU can process separate instructions. This is why I recommend at least four 64 thread work groups per CU as opposed to 256 thread work groups.
Forum: 3D Architectures & Chips 17-Mar-2013, 05:36
Replies: 144
Views: 7,656
Posted By OpenGL guy
This is not true for GCN.

This is not true for GCN.
Forum: 3D Architectures & Chips 17-Mar-2013, 02:30
Replies: 144
Views: 7,656
Posted By OpenGL guy
Not even close. A simple float add is 9 clocks...

Not even close. A simple float add is 9 clocks on GTX 680. However, a single thread can multi-issue, meaning there's only a stall if there's not enough other work to do.

I don't know about...
Forum: 3D Architectures & Chips 14-Mar-2013, 18:21
Replies: 144
Views: 7,656
Posted By OpenGL guy
It's also undesirable as it breaks any effect...

It's also undesirable as it breaks any effect that depends on the results of the previous frame(s). It also would wreak havoc with things like occlusion queries where previous frames contain the...
Forum: 3D Architectures & Chips 13-Mar-2013, 20:41
Replies: 144
Views: 7,656
Posted By OpenGL guy
This is exactly how a GPU uses registers. A...

This is exactly how a GPU uses registers. A warp/wavefront takes multiple cycles to process an instruction for all registers in use by the instruction.

CPUs use out-of-order execution and...
Forum: 3D Architectures & Chips 04-Mar-2013, 06:41
Replies: 1,507
Views: 150,053
Posted By OpenGL guy
Something seems fishy as the faster the clock...

Something seems fishy as the faster the clock speed, the larger the gain.
Forum: 3D Architectures & Chips 03-Mar-2013, 00:16
Replies: 1,507
Views: 150,053
Posted By OpenGL guy
80% CPU speed gain only results in 36%...

80% CPU speed gain only results in 36% performance, not very good CPU scaling really. As I said, you need to look at more data points. Perhaps all the benefit was going from 2.5 to 3.5GHz and the...
Forum: 3D Architectures & Chips 02-Mar-2013, 22:57
Replies: 1,507
Views: 150,053
Posted By OpenGL guy
I never said it did. What I was saying is that...

I never said it did. What I was saying is that two data points (2.5 GHz and 4.5 GHz) don't tell you a lot and I gave some ideas on how to improve the data to better understand what is happening.
Forum: 3D Architectures & Chips 02-Mar-2013, 22:17
Replies: 1,507
Views: 150,053
Posted By OpenGL guy
Two samples don't give very much data. Was the...

Two samples don't give very much data. Was the game CPU limited until 3 or 3.5 GHz or some other value? Perhaps part of your benchmark is CPU-limited and other parts are not. Was the benefit...
Forum: 3D Architectures & Chips 21-Feb-2013, 04:13
Replies: 1,292
Views: 179,503
Posted By OpenGL guy
Extra bandwidth where? And of course a load...

Extra bandwidth where? And of course a load takes up a register, where would you store the result?!

What prefetching are you referring to?

None of this makes a bit of sense to me.
Forum: 3D Architectures & Chips 21-Feb-2013, 01:14
Replies: 1,292
Views: 179,503
Posted By OpenGL guy
I suspect you mean that each CU's LDS is...

I suspect you mean that each CU's LDS is independent of the other CUs. If so, that's correct. With GCN, memory operations (local and global) are handled asynchronously from program execution. So...
Forum: 3D Architectures & Chips 19-Feb-2013, 06:35
Replies: 1,292
Views: 179,503
Posted By OpenGL guy
The handle predates 7+ years of Direct3D driver...

The handle predates 7+ years of Direct3D driver work and 3 years of OpenCL ;)

Also, bounds checking for local shared memory accesses is not new to GCN, it's been there since Evergreen (i.e. HD5XXX...
Forum: 3D Architectures & Chips 18-Feb-2013, 20:37
Replies: 1,292
Views: 179,503
Posted By OpenGL guy
Do you understand the OpenCL specification and...

Do you understand the OpenCL specification and what it says about "local" memory? Workgroups don't share local data. End of discussion.

Yes, that's what DX spec, so what? OpenCL does not say...
Showing results 1 to 25 of 500

 
Forum Jump

All times are GMT +1. The time now is 03:20.


Powered by vBulletin® Version 3.8.6
Copyright ©2000 - 2013, Jelsoft Enterprises Ltd.