SSE4+ CPU/GPU app?

[Jason00]

With the realease of the I7 processor it has SSE4.2 instructions. I was wondering if anyone was working on an optimized app for that instruction?

Or atleast the SSE4.1 instruction for the combo CPU/GPU instructions? Or is there not much difference between the SSE3 and SSE4 apps?


Thanks
Jason

[Raistmer]

With the realease of the I7 processor it has SSE4.2 instructions. I was wondering if anyone was working on an optimized app for that instruction?

Or atleast the SSE4.1 instruction for the combo CPU/GPU instructions? Or is there not much difference between the SSE3 and SSE4 apps?


Thanks
Jason

There is SSE4.1 build for AK_v8 but on some hosts it shows less performance than SSSE3 one. AFAIK there is no specific SSE4.1 made by hands there, only rebuild with compiler and libs SSE4.1 support.

ADDON: Moreover, AFAIK most of current opt app devs don't have i7 hardware available still. And it's extremely hard and ungrateful job - to optimize for target platform you can't access.
So, most probably high SSE levels will wait some more.

[Jason00]

Ok, I was just curious. Thanks for the very quick reply!! So most likely its just better to stick with the SSE3 builds. THanks

[Jason G]

At this stage it appears, due to radical architectural differences & embryonic state of i7 related SDKs & documentation, that the SSE3  (That's two S's, not three) may perform  better on i7 than either SSSE3x or SSE4.1 builds. As they say "Your mileage may vary".

Give it a go 

[Jason00]

So would it just be a matter of modifying the app_info.xml file with the name of the SSE4.1 file and of course pasting that file in the Seti directory?


***I"m using the "Raistmer's_opt_pack_V10a_x64_SSE3_Intel_MB_CPU_GPU_team" package at the moment.****

[Jason G]

Sure! making sure to get the application name exactly right.  Do remember to try the SSE3 one though, and compare different builds.  SSE3, SSSE3x & SSE4.1 could all be worth a try depending on your system ... (So many variables, Chips, OC, Cache, RAM, Mobo chipset, OS) .. too many for us poor kids to test with no i7's at all 

[Jason00]

Sure! making sure to get the application name exactly right.  Do remember to try the SSE3 one though, and compare different builds.  SSE3, SSSE3x & SSE4.1 could all be worth a try depending on your system ... (So many variables, Chips, OC, Cache, RAM, Mobo chipset, OS) .. too many for us poor kids to test with no i7's at all 

My I7 build was my first new computer since I got my AMD Athalon 64 3200 when it first came out, what 4 to 5 years ago. It was time for a much needed upgrade.


*** Would I also need to copy over the  .manifest file, and the .pdb files too?***

[Jason G]

My I7 build was my first new computer since I got my AMD Athalon 64 3200 when it first came out, what 4 to 5 years ago. It was time for a much needed upgrade.


*** Would I also need to copy over the  .manifest file, and the .pdb files too?***

Those files are 'optional extras'... nice to have them if you crash though.  That's one 'quantum leap' alright! 

[Jason00]

Ok so I discovered it's not as simple as I thought. Since I wanted to use the SSE4.1 build with the V10a CPU/GPU combo, I just renamed the app in the app_info.xml file. BUT, all I got was now ten versions of Win64_AK_v8_SSE4.1 app running. Instead of 10 cpu apps and the two gpu cuda apps.

I'm guessing there's more to the CPU/GPU team of file than just replacing files the folder and app_info.xml file.

So did I either do something wrong or is there more compiling of the files that would have to be done to get it to work?

Here's the app_info.xml that I modified.



<app_info>
    <app>
        <name>astropulse</name>
    </app>
    <file_info>
        <name>ap_5.00r103_SSE3.exe</name>
        <executable/>
    </file_info>
    <app_version>
        <app_name>astropulse</app_name>
        <version_num>500</version_num>
        <file_ref>
            <file_name>ap_5.00r103_SSE3.exe</file_name>
            <main_program/>
        </file_ref>
    </app_version>
   <app>
      <name>astropulse_v5</name>
   </app>
   <file_info>
           <name>ap_5.03r112_SSE3.exe</name>
           <executable/>
       </file_info>
       <app_version>
           <app_name>astropulse_v5</app_name>
           <version_num>503</version_num>
           <file_ref>
                  <file_name>ap_5.03r112_SSE3.exe</file_name>
                  <main_program/>
           </file_ref>
       </app_version>
<app>
<name>setiathome_enhanced</name>
</app>
<file_info>
<name>AK_v8_win_x64_SSE41.exe</name>
<executable/>
</file_info>
<file_info>
<name>cudart.dll</name>
<executable/>
</file_info>
<file_info>
<name>cufft.dll</name>
<executable/>
</file_info>
<file_info>
<name>libfftw3f-3-1-1a_upx.dll</name>
<executable/>
</file_info>
<file_info>
<name>MB_6.08_mod_CUDA_V10.exe</name>
<executable/>
</file_info>

<app_version>
<app_name>setiathome_enhanced</app_name>
<version_num>528</version_num>
<file_ref>
<file_name>AK_v8_win_x64_SSE41.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>cudart.dll</file_name>
</file_ref>
<file_ref>
<file_name>cufft.dll</file_name>
</file_ref>
<file_ref>
<file_name>libfftw3f-3-1-1a_upx.dll</file_name>
</file_ref>
<file_ref>
<file_name>MB_6.08_mod_CUDA_V10.exe</file_name>
</file_ref>
</app_version>

<app_version>
<app_name>setiathome_enhanced</app_name>
<version_num>603</version_num>
<file_ref>
<file_name>AK_v8_win_x64_SSE41.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>cudart.dll</file_name>
</file_ref>
<file_ref>
<file_name>cufft.dll</file_name>
</file_ref>
<file_ref>
<file_name>libfftw3f-3-1-1a_upx.dll</file_name>
</file_ref>
<file_ref>
<file_name>MB_6.08_mod_CUDA_V10.exe</file_name>
</file_ref>
</app_version>


<app_version>
<app_name>setiathome_enhanced</app_name>
<version_num>607</version_num>
<file_ref>
<file_name>AK_v8_win_x64_SSE41.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>cudart.dll</file_name>
</file_ref>
<file_ref>
<file_name>cufft.dll</file_name>
</file_ref>
<file_ref>
<file_name>libfftw3f-3-1-1a_upx.dll</file_name>
</file_ref>
<file_ref>
<file_name>MB_6.08_mod_CUDA_V10.exe</file_name>
</file_ref>
</app_version>

<app_version>
<app_name>setiathome_enhanced</app_name>
<version_num>608</version_num>
<file_ref>
<file_name>AK_v8_win_x64_SSE41.exe</file_name>
<main_program/>
</file_ref>
<file_ref>
<file_name>cudart.dll</file_name>
</file_ref>
<file_ref>
<file_name>cufft.dll</file_name>
</file_ref>
<file_ref>
<file_name>libfftw3f-3-1-1a_upx.dll</file_name>
</file_ref>
<file_ref>
<file_name>MB_6.08_mod_CUDA_V10.exe</file_name>
</file_ref>
</app_version>

</app_info>

[Jason G]

For 'team' functionality, you would need to use whatever equivalent 'team' builds raistmer has made. Sorry I diidn't notice you were asking about GPU stuff  (even though obviosuly in the thread title  ). Please treat 'team builds' as distinct entities from 'Lunatics  Windows port of AKv8'.  AFAIK there are no SSE4.1 builds in Raistmer's set, because they run 1% slower than SSSE3x on his quad.

Jason

[Jason00]

Ok cool, thanks for the info. Guess I'll continue doing what I do best then. Just download all the great stuff you guys create and crunch away!!!

[Raistmer]

So would it just be a matter of modifying the app_info.xml file with the name of the SSE4.1 file and of course pasting that file in the Seti directory?


***I"m using the "Raistmer's_opt_pack_V10a_x64_SSE3_Intel_MB_CPU_GPU_team" package at the moment.****

SUre not
"team" mod called mod because it used MODIFIED AK_v8 files BTW
There is no SSE4.1 team build available.
But you still can test SSE4.1 AK_v8 in standalone runs (but not as member of GPU/CPU team).

[Raistmer]

Ok so I discovered it's not as simple as I thought. Since I wanted to use the SSE4.1 build with the V10a CPU/GPU combo, I just renamed the app in the app_info.xml file. BUT, all I got was now ten versions of Win64_AK_v8_SSE4.1 app running. Instead of 10 cpu apps and the two gpu cuda apps.

It seems you learned what "mod" means by experiment - the best way to learn indeed

[Jason00]

If you guys don't have an I7 processor to test with, I'd be a willing volunteer! Just let me know.

Here's my setup.

MB – Asus Rampage 2 Extreme
CPU – Intel Core i7 920 CPU @  w/Apogee GTZ WB
MB – Asus Rampage II Extreme Bios Ver. 1104
RAM – OCZ Platinum 6GB (3 x 2GB) 240-Pin DDR3 SDRAM DDR3 1333 (PC3 10666)
VGA – SLI – EVGA GeForce GTX 260 Core 216 896MB 448-bit GDDR3 w/D-Tek FuZion GFX 2 & UNI-Sink Heatskink
PSU – Corsair HX1000 Power Supply
HDD – Raid 0 – Western Digital Cavier Black 500GB 7,200 32MB
WATER COOLING – x2 MRC320 Radiators, Swiftech MCP655 Pump, 8 Scythe SY1225SL12SH 120mm "Slipstream" Fans
OTHERS – 1 WD Black 1TB HD, 2 DVD Burners, & 1 Flash card reader

[Jason G]

Cheers!... It's really more of a case of having an i7 to build with though, and Intel pulling their collective fingers out for  us "Green Belt Intel Software Development Partners" to place more than one geriatric nerd on the SDK help desk, and put more than marketing diatribe in there. ("I7 is faster".. yeah that's helpful)

Sorry, attended a webinar last week, and figure most of the software development tm is contracted to Russia and China.  So we could be waiting some time for updates to the "optimisation reference manual" for starters.

Jason