Jump to content

Wrong readings on GeForce GTX Texel Fillrate & System Swap & VM


javanse

Recommended Posts

Hi,

It looks like AIDA64 uses the wrong formula for calculating the Texel Fillrate for GeForce GTX 400 GPUs, while Pixel Fillrate is calculated correctly.

Pixel Fillrate = Core Clock * ROP Count

My GTX 470 has 40 ROPs and is clocked @ 750 MHz => 30 GPixel/s, that value is correctly shown by AIDA and GPU-Z.

Texel Fillrate = Core Clock * TMU Count

My GTX 470 has 56 TMUs and is clocked @ 750 MHz => 42 GTexel/s, that value is correctly shown by GPU-Z while AIDA64 shows exactly 4 times as much, 168 GTexel/s.

FYI, this calculation error is visible in all P-States.

Second thing, i can't understand some of the readings on the Mainboard -> Memory page.

I've 8 GB RAM with a fixed 2 GB min / 2 GB max pagefile which AIDA shows correctly.

But it calculates a swap space of 10 GB, apparently RAM+Swap, and 18 GB virtual memory, apparently RAM+swap space, which doesn't make any sense to me.

Shouldn't it read a swap space of 2 GB and 10 GB virtual memory, as confirmed by the 10 GB commit charge limit reading from Process Explorer from Sysinternals?

Link to comment
Share on other sites

We've fixed the texture fillrate calculation in the latest beta version of AIDA64 Extreme Edition available at:

http://www.aida64.com/downloads/aida64extremebuild1243bw4ylkfd0mzip

After upgrading to this new version, make sure to restart Windows to finalize the upgrade.

As for the memory sizes, they can be confusing, but that's how Windows manages swap space and paging files. Even if you disable paging files completely, Windows will still manage some swap space. Quite hard to follow that logic, but that's just how Windows works :)

Link to comment
Share on other sites

Thanks for the update Fiery, Texel Fillrate reading is now correct :)

But there's another issue related to the TMUs:

For my GTX 470, AIDA reads 40 Pixel Pipelines (aka ROPs), which is correct, but also "TMU Per Pipeline = 1", which is wrong in two ways at once.

First, it implies 40 while there are 56 TMUs. And second, the item name "TMU Per Pipeline" itself is only correct up to G71. Since G80, ROPs and TMUs are decoupled from each other, so IMHO it should be renamed to "Texture Mapping Units" only as the amount of ROPs is related to the bus width of the memory interface while the amount of TMUs is related to the streaming multiprocessors.

Amount of TMUs: 448 Shader / 32 Shader per SM gives 14 SM. Each SM has 4 TMUs for a total of 56 TMUs.

Amount of ROPs: 320 bit MI / 64 bit width per ROP-Partition gives 5 ROP-P. Each ROP-P has 8 ROPs for a total of 40 ROPs.

kthxbye and a have a nice weekend :)

Link to comment
Share on other sites

We've fixed the TMU issue, and also implemented TMU count detection and texel fillrate calculation for R600 and later GPUs by ATI/AMD.

Please upgrade to the latest beta version of AIDA64 Extreme Edition available at:

http://www.aida64.com/downloads/aida64extremebuild1251oq1xi5ergvzip

After upgrading to this new version, make sure to restart Windows to finalize the upgrade.

Let me know how it works.

Thanks,

Fiery

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.



×
×
  • Create New...