2444 replies to this topic

[spakk]

I read this post ( http://software.inte...ms/topic/305798 ) that led me to the following link.

Intel® Processor Identification and the CPUID Instruction

http://213.174.154.6...c=24161831 .pdf


Intel® Compiler Options for Intel® SSE and Intel® AVX generation (SSE2, SSE3, SSE3_ATOM, SSSE3, SSE4.1, SSE4.2, AVX, AVX2) and processor-specific optimizations  

http://software.inte...436?wapkw=ssse3


......Which processor is targeted by default?

• On IA-32 systems running Windows*, /arch:SSE2 is on by default. On IA-32 systems running Linux*, -msse2 is on by default. The resulting code path should run on the Intel Pentium 4 and Intel Xeon processors with SSE2 support and other later Intel processors or compatible non-Intel processors with SSE2 support.
  
• On IA-32 systems running Mac OS* X, -xSSE3 is on by default. The compiler may generate SSE3, SSE2 and SSE instructions and the code is optimized for enhanced Pentium M processor microarchitecture.
  
• On Intel 64 systems running Mac OS* X, -xSSSE3 is on by default. The compiler may generate SSSE3, SSE3, SSE2 and SSE instructions and the code is optimized for the Intel® Core™ microarchitecture.

To target older IA-32 systems without support for SSE2 instructions, such as systems based on the Intel® Pentium® III Processor, use the switch /arch:ia32 (Windows*) or -mia32 (Linux*).

For information about other, older processor targeting options and their relation to the recommended options above, see
http://software.inte...geting-overview

Edited by spakk, 11 November 2012 - 03:42 PM.

[spakk]

This is a hot potato that we have ahead of us.

[SS01]

JaS, on 27 July 2011 - 01:02 AM, said:

This is the same debate as we had with sse2 cpu's back in the day. The emualtor for sse3 was able to be made because the sse3 calls were able to be broken down in to their multiple sse2 instruction counterparts iirc. ssse3 afaik can't be emulated in the same way.

Maybe creating an SSSE3 emu really is impossible? in which case, we need to work on FX/Trinity/Piledriver.

[spakk]

SS01, on 11 November 2012 - 11:11 PM, said:

Maybe creating an SSSE3 emu really is impossible? in which case, we need to work on FX/Trinity/Piledriver.

Please see the folder content of Bochs emulator in detail to accurately and check that is possible to take the code snippets. I started with it, but with my little knowledge of it can be very, very long daueren until I am finished. To compare the data I use ExamDiff Pro.

spakk, on 11 November 2012 - 11:43 PM, said:

Please see the folder content of Bochs emulator in detail to accurately and check that is possible to take the code snippets. I started with it, but with my little knowledge of it can be very, very long daueren until I am finished. To compare the data I use ExamDiff Pro.

we not need reinvent the wheel again

[ZackehSoul]

SS01, on 11 November 2012 - 11:11 PM, said:

Maybe creating an SSSE3 emu really is impossible? in which case, we need to work on FX/Trinity/Piledriver.

Nothing is impossible, just too difficult and unrealistic for people to achieve

[SS01]

irc.osx86.hu is down again, and likely for good, or at least a very long time. let's move to #LegacyKernel on irc.systemnet.info

UPDATE: it's back. What the hell..

[spakk]

SS01, on 12 November 2012 - 12:39 PM, said:

irc.osx86.hu is down again, and likely for good, or at least a very long time. let's move to #LegacyKernel on irc.systemnet.info

UPDATE: it's back. What the hell..

is On my friend

[spakk]

ZackehSoul, on 12 November 2012 - 02:30 AM, said:

Nothing is impossible, just too difficult and unrealistic for people to achieve

We must not sit in the corner and cry. This brings us not further, we should  the good examples of the others understand and implement for our Project.

[Deltac0]

Why wouldn't SSSE3 emulator be possible? As far as I know, the SSSE3 instructions can be performed with multiple SSE3 and other instructions.

spakk, on 11 November 2012 - 11:45 PM, said:

Please see the folder content of Bochs emulator in detail to accurately and check that is possible to take the code snippets. I started with it, but with my little knowledge of it can be very, very long daueren until I am finished. To compare the data I use ExamDiff Pro.



we not need reinvent the wheel again

I'm not sure what you're trying to proof with that PCSX2 video. PCSX2 is a PS2 emulator for PC, which has this plugin called GSDX, which comes in SSE2, SSSE3 and SSSE4 versions.
There is no SSSE3 emulator on PCSX2. Or are you trying to compare the differences between SSE2 and SSSE3 versions of GSDX?
This is pretty complicated...

http://wiki.pcsx2.ne....php?title=GSDX

[spakk]

Deltac0, on 12 November 2012 - 08:19 PM, said:

Why wouldn't SSSE3 emulator be possible? As far as I know, the SSSE3 instructions can be performed with multiple SSE3 and other instructions.




I'm not sure what you're trying to proof with that PCSX2 video. PCSX2 is a PS2 emulator for PC, which has this plugin called GSDX, which comes in SSE2, SSSE3 and SSSE4 versions.
There is no SSSE3 emulator on PCSX2. Or are you trying to compare the differences between SSE2 and SSSE3 versions of GSDX?
This is pretty complicated...

http://wiki.pcsx2.ne....php?title=GSDX

You're right, I wanted to create a little momentum here in the hut.

[SS01]

you know spakk, you're right. We must keep trying! Unfortunately, I have no knowledge on how to write an SSSE3 emulator. However, I will continue to work on the kernel.

[spakk]

spakk, on 12 November 2012 - 08:50 PM, said:

You're right, I wanted to create a little momentum here in the hut.

I have searched the Internet and found a contribution to the "x86 open64 compiler suite for x86 Open64 version 4.5.2 ".
The other link is the result "SSSE3" on the AMD-website

x86 open64 compiler suite:
http://developer.amd...compiler-suite/   ***{censored}*  only a Linux Executable version  

here the x86 Open64 User’s Guide [PDF]
www.workupload.com/file/9wGzcsI

other link, Open64 community

http://www.open64.net/home.html

result "SSSE3" on the AMD-website:

** On the AMD website you have to register your to read the contributions

Edited by spakk, 12 November 2012 - 10:35 PM.

[spakk]

SS01, on 12 November 2012 - 09:07 PM, said:

you know spakk, you're right. We must keep trying! Unfortunately, I have no knowledge on how to write an SSSE3 emulator. However, I will continue to work on the kernel.

theconnatic has me recently written that he is going to write an SSSE3 emulator . I hope I have understood him.

[gils83]

)-

Edited by gils83, 08 December 2012 - 12:04 AM.

[Deltac0]

gils83, on 13 November 2012 - 12:49 PM, said:

Bonjour à tous ,

je suis nouveau sur insanelymac , j'éspère vous aider sur ce nouveau défit pour le kernel 12.0 pour processeur AMD . Je possède une plateforme AMD FX et n'hésitez pas à me contacter pour d'éventuelles essaies , je suis un peu là pour ça .



Hi all,  

I'm new to InsanelyMac, I hope to help you on this new challenge for the 12.0 kernel for AMD. I have an AMD FX platform and do not hesitate to contact me for any try, I'm here for a bit.


PS :  Thank you to tell me if my English is correct?

In my opinion, your English is pretty damn correct, at least better than mine.

I'm just trying to setup my other AMD rig with Athlon64 4000+ @ 2.5Ghz and 1.5GB DDR, 4 IDE HDDS (80GBS each, was thinking a raid-0 array, would be awesome )
It's now running 10.6.8 and just got my wlan working. Now I need XCode and Lion on other HDD to test my coding skills.
Previously (2 months ago), I managed to compile a working Lion kernel for AMD (32-bit) by myself, with a little help of .diff files...
I'm just curious about this SSSE3 emulation, I think I could find some source code of programs using SSSE3 and SSE3 and compare the differences.

[gils83]

Deltac0, on 13 November 2012 - 03:07 PM, said:

In my opinion, your English is pretty damn correct, at least better than mine.

I'm just trying to setup my other AMD rig with Athlon64 4000+ @ 2.5Ghz and 1.5GB DDR, 4 IDE HDDS (80GBS each, was thinking a raid-0 array, would be awesome )
It's now running 10.6.8 and just got my wlan working. Now I need XCode and Lion on other HDD to test my coding skills.
Previously (2 months ago), I managed to compile a working Lion kernel for AMD (32-bit) by myself, with a little help of .diff files...
I'm just curious about this SSSE3 emulation, I think I could find some source code of programs using SSSE3 and SSE3 and compare the differences.

I using Google translation, it is not the best solution but I'm trying to find a French not too colorful for google translation does not tell too much stupidity.    

there Bronzovka aka DmitriK and RAWX86 successful work on this kernel 11.4, there is no way he will adapt their work on the 12.0 kernel ?



je continu mes tests pour SS01    

[spakk]

@ gils83 ,  @Deltac0

Welcome to the club

@Deltac0
see my Post #400 and invite you the Mac and Windows version and check the folders and code snippets in detail and tell me what you think of it.

[gils83]

gils83, on 13 November 2012 - 03:31 PM, said:

I using Google translation, it is not the best solution but I'm trying to find a French not too colorful for google translation does not tell too much stupidity.    

there Bronzovka aka DmitriK and RAWX86 successful work on this kernel 11.4, there is no way he will adapt their work on the 12.0 kernel ?



je continu mes tests pour SS01

I tried following the instructions on the SS01 base ML 10.8 DP1 (best result for the core2duo)

But unfortunately, there are no way to block the AMD FX, it restarts without any following message.

I test on a core2duo and a FX 6100 with the additional flag "busratio = 165"

you approach!

d'autres idées ?  







test core2duo



’<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com...yList-1.0.dtd">
<plist version="1.0">
<dict>
<key>kernel</key>
<string>amd_kernel_beta</string>
<key>Kernel Flags</key>
<string>arch=x86_64 maxmem=4096 amd_kernel_beta PCIRootUID=1 npci=0x3000 pci=off -v -f</string>
<key>Theme</key>
<string>default</string>
<key>EthernetBuiltIn</key>
<string>Yes</string>
<key>GraphicsEnabler</key>
<string></string>
</dict>
</plist>




<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com...yList-1.0.dtd">
<plist version="1.0">
<dict>
<key>SMbiosvendor</key>
<string>AppleInc.</string>
<key>SMbiosversion</key>
<string>MP31.88Z.006C.B05.0802291410</string>
<key>SMboardproduct</key>
<string>Mac-F42C88C8</string>
<key>SMfamily</key>
<string>MacPro</string>
<key>SMmanufacturer</key>
<string>AppleInc.</string>
<key>SMproductname</key>
<string>MacPro3,1</string>
<key>SMserial</key>
<string>G8135543XYL</string>
</dict>
</plist>

spakk, on 13 November 2012 - 04:41 PM, said:

@ gils83 ,  @Deltac0

Welcome to the club

@Deltac0
see my Post #400 and invite you the Mac and Windows version and check the folders and code snippets in detail and tell me what you think of it.

thank spakk  

[Deltac0]

spakk, on 13 November 2012 - 04:41 PM, said:

@ gils83 ,  @Deltac0

Welcome to the club

@Deltac0
see my Post #400 and invite you the Mac and Windows version and check the folders and code snippets in detail and tell me what you think of it.

Thanks! I'm a bit busy setting up my 10.6 atm, but will definately check it out!
And btw, I may have a debug method for the FX random restart.
Compile a new kernel, with messages like "DEBUG #1", "DEBUG #2", "DEBUG #3" showing up between different stages of boot. Then if you can see "DEBUG #2", but not "#DEBUG #3" before the restart, you know the problem is somewhere between those messages. I know adding the messages just like that isn't very easy, but that could help.

[gils83]

I have tried many flag but I can not block the restart.  

I manage to get a message with a 4800 + !!! this is frustrating !!!