Trying to find the cause for JITC errors.
Moderators: Cat_7, Ronald P. Regensburg, ClockWise
Trying to find the cause for JITC errors.
If your box can run SheepShaver w/ JIT enabled, without getting errors or hangage, then can you take a few seconds to post what ROM, CPU, and Operating System you're running (If Windows, then what version and SP)?
Gotta see what's causing this. Because from what I see, it works for some people, and not others.
Gotta see what's causing this. Because from what I see, it works for some people, and not others.
I can run SS with jit enabled + extensions on for a while (+/- 30 mins) before it freezes.
ROM - MacOS ROM update 1.0
CPU - Athlon 64 3200+ (S939)
Host OS - Windows XP Pro SP2 + all updates
Client OS - MacOS 8.6 (installed 8.5 and updated to 8.6)
On SuSE Linux 9.2 AMD64 SS runs flawless.
Edit:
My CPU flags are:
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmovpat pse36 clflush mmx fxsr sse sse2 syscall nx mmxext lm 3dnowext 3dnow
ROM - MacOS ROM update 1.0
CPU - Athlon 64 3200+ (S939)
Host OS - Windows XP Pro SP2 + all updates
Client OS - MacOS 8.6 (installed 8.5 and updated to 8.6)
On SuSE Linux 9.2 AMD64 SS runs flawless.
Edit:
My CPU flags are:
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmovpat pse36 clflush mmx fxsr sse sse2 syscall nx mmxext lm 3dnowext 3dnow
Last edited by Jack Hair on Thu Mar 03, 2005 6:53 am, edited 1 time in total.
- PPC_Digger
- Forum All-Star
- Posts: 1050
- Joined: Thu Jul 22, 2004 9:02 am
- Location: Israel
-
- Apple Corer
- Posts: 271
- Joined: Mon Sep 23, 2002 6:53 am
At this point JIt seems hit and miss.
I wonder why? Maybe CPU makers use different cores even among same family of chips or something? It seems likely that CPU is the blame for this, although Linux doesn't care what CPU is used, so maybe it's just that Windows lacks certian features that Linux has to overcome CPU differences.
I wonder why? Maybe CPU makers use different cores even among same family of chips or something? It seems likely that CPU is the blame for this, although Linux doesn't care what CPU is used, so maybe it's just that Windows lacks certian features that Linux has to overcome CPU differences.
The AltiVec emulation code uses MMX or SSE optimizations were suitable. I was using gcc intrinsincs, not direct assembly code. Cygwin gcc version used was 3.3 something, that branch had known bugs in the intrinsics code (like sometimes swapped args). Various distributors, including Mandrakelinux I use, had a correctly patched version of gcc already. Now, to validate this hopthesis, people can also report their CPU capabilities. Something like:The Balance Of Judgement wrote:It seems likely that CPU is the blame for this, although Linux doesn't care what CPU is used, so maybe it's just that Windows lacks certian features that Linux has to overcome CPU differences.
[gb@nocona gb]$ grep ^flags /proc/cpuinfo
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm syscall lm pni monitor ds_cpl cid
under linux. I don't know for Windows, you might probably need tools like CPU-Z or cpuinfo, whatever tells you your CPU supports MMX, SSE, SSE2, etc.
ok so dont know if it may help but here is my CPU-Z dump
----------------------
CPU-Z version 1.27.1
----------------------
CPUID Output
------------------------------------------------------------------------------
Number of CPUs 1
Name AMD Athlon XP
Code name Thoroughbred
Specification AMD Athlon(tm)
Family/Model/Stepping 681
Extended Family/Model 7/8
Package Socket A
Core Stepping B0
Technology 0.13µ
Instructions Sets MMX, Extended MMX, 3DNow!, Extended 3DNow!, SSE
Clock Speed 2004.6 MHz
Clock multiplier x10.0
Front Side Bus Frequency 200.5 MHz
Bus Speed 400.9 MHz
L1 Data Cache 64 KBytes, 2-way set associative, 64 Bytes line size
L1 Instruction Cache 64 KBytes, 2-way set associative, 64 Bytes line size
L2 Cache 256 KBytes, 16-way set associative, 64 Bytes line size
L2 Speed 2004.6 MHz (Full)
L2 Location On Chip
L2 Data Prefetch Logic yes
L2 Bus Width 64 bits
Software
------------------------------------------------------------------------------
Windows Version Microsoft Windows XP Professional Service Pack 2 (Build 2600)
DirectX Version 9.0c
----------------------
CPU-Z version 1.27.1
----------------------
CPUID Output
------------------------------------------------------------------------------
Number of CPUs 1
Name AMD Athlon XP
Code name Thoroughbred
Specification AMD Athlon(tm)
Family/Model/Stepping 681
Extended Family/Model 7/8
Package Socket A
Core Stepping B0
Technology 0.13µ
Instructions Sets MMX, Extended MMX, 3DNow!, Extended 3DNow!, SSE
Clock Speed 2004.6 MHz
Clock multiplier x10.0
Front Side Bus Frequency 200.5 MHz
Bus Speed 400.9 MHz
L1 Data Cache 64 KBytes, 2-way set associative, 64 Bytes line size
L1 Instruction Cache 64 KBytes, 2-way set associative, 64 Bytes line size
L2 Cache 256 KBytes, 16-way set associative, 64 Bytes line size
L2 Speed 2004.6 MHz (Full)
L2 Location On Chip
L2 Data Prefetch Logic yes
L2 Bus Width 64 bits
Software
------------------------------------------------------------------------------
Windows Version Microsoft Windows XP Professional Service Pack 2 (Build 2600)
DirectX Version 9.0c
Last edited by ataxy on Fri Mar 04, 2005 6:12 pm, edited 1 time in total.
- PPC_Digger
- Forum All-Star
- Posts: 1050
- Joined: Thu Jul 22, 2004 9:02 am
- Location: Israel
Just so you'd know, on my Athlon-4 (which SS runs flawlessly on with jit on both linux and windows) the following are supported: MMX, Extended MMX, 3DNow!, 3DNow!+ and SSE (no SSE2).
Here are the flags:
flags : fpu vme de pse tsc msr pae mce cx8 sep mtrr pge mca cmov pat pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow
As far as I can see right now, only Athlon-4 and Athlon 64 work fine. We need someone with an Athlon XP and Linux to tell us about their expirience.
Here are the flags:
flags : fpu vme de pse tsc msr pae mce cx8 sep mtrr pge mca cmov pat pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow
As far as I can see right now, only Athlon-4 and Athlon 64 work fine. We need someone with an Athlon XP and Linux to tell us about their expirience.
cpu specs
my athlon xp 3200 fails with jitc under win2k sp4. CPUZ below.
Obviously would be great if you could figure this out gb.
Number of CPUs 1
Name AMD Athlon XP
Code name Barton
Specification AMD Athlon(tm) XP 3200+
Family/Model/Stepping 6A0
Extended Family/Model 7/A
Package Socket A
Technology 0.13µ
Instructions Sets MMX, Extended MMX, 3DNow!, Extended 3DNow!, SSE
Clock Speed 2205.1 MHz
Clock multiplier x11.0
Front Side Bus Frequency 200.5 MHz
Bus Speed 400.9 MHz
P-Rating 3200+
L1 Data Cache 64 KBytes, 2-way set associative, 64 Bytes line size
L1 Instruction Cache 64 KBytes, 2-way set associative, 64 Bytes line size
L2 Cache 512 KBytes, 16-way set associative, 64 Bytes line size
L2 Speed 2205.1 MHz (Full)
L2 Location On Chip
L2 Data Prefetch Logic yes
L2 Bus Width 64 bits
Obviously would be great if you could figure this out gb.
Number of CPUs 1
Name AMD Athlon XP
Code name Barton
Specification AMD Athlon(tm) XP 3200+
Family/Model/Stepping 6A0
Extended Family/Model 7/A
Package Socket A
Technology 0.13µ
Instructions Sets MMX, Extended MMX, 3DNow!, Extended 3DNow!, SSE
Clock Speed 2205.1 MHz
Clock multiplier x11.0
Front Side Bus Frequency 200.5 MHz
Bus Speed 400.9 MHz
P-Rating 3200+
L1 Data Cache 64 KBytes, 2-way set associative, 64 Bytes line size
L1 Instruction Cache 64 KBytes, 2-way set associative, 64 Bytes line size
L2 Cache 512 KBytes, 16-way set associative, 64 Bytes line size
L2 Speed 2205.1 MHz (Full)
L2 Location On Chip
L2 Data Prefetch Logic yes
L2 Bus Width 64 bits
-
- Space Cadet
- Posts: 1
- Joined: Fri Mar 04, 2005 4:16 am
- Location: Knoxville, TN USA
I use an Athlon XP 2500+, which has the same specs as a 3200+, but a clock rate of 1.83 GHz and FSB of 166 MHz.
SS runs JIT without crashing in Mandrake Linux 10.1, but it is very squirrely. It speeds up, then slows down again at seemingly random intervals. Switching off JIT resulted in a major slowdown, but a more consistent experience. I used the precompiled SS 2.2 RPM. (I couldn't seem to install the packages required to compile it myself.) Hope this helps, Gwenole. :D
SS runs JIT without crashing in Mandrake Linux 10.1, but it is very squirrely. It speeds up, then slows down again at seemingly random intervals. Switching off JIT resulted in a major slowdown, but a more consistent experience. I used the precompiled SS 2.2 RPM. (I couldn't seem to install the packages required to compile it myself.) Hope this helps, Gwenole. :D
I used 3.3 on gentoo, maybe that's why it (JIT) won't work for me. :Sgb wrote:The AltiVec emulation code uses MMX or SSE optimizations were suitable. I was using gcc intrinsincs, not direct assembly code. Cygwin gcc version used was 3.3 something, that branch had known bugs in the intrinsics code (like sometimes swapped args). Various distributors, including Mandrakelinux I use, had a correctly patched version of gcc already. Now, to validate this hopthesis, people can also report their CPU capabilities. Something like:The Balance Of Judgement wrote:It seems likely that CPU is the blame for this, although Linux doesn't care what CPU is used, so maybe it's just that Windows lacks certian features that Linux has to overcome CPU differences.
[gb@nocona gb]$ grep ^flags /proc/cpuinfo
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm syscall lm pni monitor ds_cpl cid
under linux. I don't know for Windows, you might probably need tools like CPU-Z or cpuinfo, whatever tells you your CPU supports MMX, SSE, SSE2, etc.
- PPC_Digger
- Forum All-Star
- Posts: 1050
- Joined: Thu Jul 22, 2004 9:02 am
- Location: Israel
Right now it looks like the problem is Athlon XP with windows, and certain linux distros.LimeMaverick wrote:I use an Athlon XP 2500+, which has the same specs as a 3200+, but a clock rate of 1.83 GHz and FSB of 166 MHz.
SS runs JIT without crashing in Mandrake Linux 10.1, but it is very squirrely. It speeds up, then slows down again at seemingly random intervals. Switching off JIT resulted in a major slowdown, but a more consistent experience. I used the precompiled SS 2.2 RPM. (I couldn't seem to install the packages required to compile it myself.) Hope this helps, Gwenole. :D