mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Software (https://www.mersenneforum.org/forumdisplay.php?f=10)
-   -   P4 SSE2 not present in Options->CPU :-( (https://www.mersenneforum.org/showthread.php?t=94)

paulunderwood 2002-09-15 17:15

P4 SSE2 not present in Options->CPU :-(
 
I am running 22.9.1 on a P4 under Window 98SE and when I inspect Options->CPU I see: CPU features: RDTSC, CMOV, PREFETCH, MMX, SSE ... but no SSE2 -- like there is in the example in the help topics on Options->CPU. The machine is running not much quicker than an Athlon XP 1600+.

I am using Prime95 to find factors of 2^64680-1.

Paul

Prime95 2002-09-15 18:37

You can try setting "CpuSupportsSSE2=1" in local.ini. However, I suspect this cause a crash. Prime95 most likely is not displaying the SSE2 setting because CPUID reports that SSE2 is supported, but attempting to execute an SSE2 instruction raises an exception.

Windows 95 operating system will not support SSE2 applications. I thought all Win98 versions did, but maybe not. If it is the OS, then upgrading the OS is the only solution (unless someone knows about a registry entry you can set to enable SSE2 support).

paulunderwood 2002-09-16 02:08

(SSE does not appear on my XP1600+ but it does on my XP1700+.)

I added CpuSupportsSSE2=1 and rebooted and SSE2 appeared in Options->CPU. I ran the benchtest and both XP1600+ and P4 are true to form. But for factoring they are running at equal speeds!

Paul

outlnder 2002-09-16 02:28

Athlons, XPs and Durons do NOT have SSE2 optimazations. Only P4s do. XPs have SSE which is not SSE2. Sorry.

paulunderwood 2002-09-16 02:33

I know!

XP1700+ looks okay
XP1600+ say no SSE
P4 now reporting SSE2 with CpuSupportsSSE2=1 but running like XP1600+

Paul

Prime95 2002-09-16 03:36

Can you send me the local.ini file and a copy of what Options/CPU is reporting (you can probably get much of this written to results.txt by doing Options/Benchmark).

Prime95 2002-09-16 03:37

Also, have you tried setting CpuSupportsPrefetch=1 and CpuSupportsSSE=1 in local.ini?

paulunderwood 2002-09-16 13:25

P4
==
local.ini:

CPUType=12
CPUSpeed=2540
CPUHours=24
DayMemory=228
NightMemory=228
DayStartTime=450
DayEndTime=451
RollingAverage=1000
RollingStartTime=0
OldCpuType=12
OldCpuSpeed=2550
RunOnBattery=1
CpuSupportsSSE2=1

results.txt:

Intel(R) Pentium(R) 4 CPU 2.40GHz
CPU speed: 2548.77 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX, SSE, SSE2
L1 cache size: 8 KB
L2 cache size: 512 KB
L1 cache line size: 64 bytes
L2 cache line size: 64 bytes
TLBS: 64
Prime95 version 22.9, RdtscTiming=1
Best time for 256K FFT length: 10.625 ms.
...


XP1600+
=======

local.ini:

OldCpuType=11
OldCpuSpeed=1392
DayMemory=63
NightMemory=63
CPUHours=24
DayStartTime=450
DayEndTime=451
RunOnBattery=1

results.txt:

AMD Athlon(tm) XP 1600+
CPU speed: 1391.66 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX
L1 cache size: 64 KB
L2 cache size: 256 KB
L1 cache line size: 64 bytes
L2 cache line size: 64 bytes
L1 TLBS: 32
L2 TLBS: 256
Prime95 version 22.9, RdtscTiming=1
Best time for 256K FFT length: 28.769 ms.
...


XP1700+ (okay):
===============

local.ini:

CPUType=11
CPUSpeed=1461
RollingAverage=1000
RollingStartTime=0
CPUHours=24
DayMemory=32
NightMemory=32
DayStartTime=450
DayEndTime=451
RunOnBattery=1
OldCpuType=11
OldCpuSpeed=1460

results.txt:

AMD Athlon(tm) XP 1700+
CPU speed: 1461.31 MHz
CPU features: RDTSC, CMOV, PREFETCH, MMX, SSE
L1 cache size: 64 KB
L2 cache size: 256 KB
L1 cache line size: 64 bytes
L2 cache line size: 64 bytes
L1 TLBS: 32
L2 TLBS: 256
Prime95 version 22.9, RdtscTiming=1
Best time for 256K FFT length: 29.724 ms.

HTH

Prime95 2002-09-16 14:18

[quote="paulunderwood"]I ran the benchtest and both XP1600+ and P4 are true to form. But for factoring they are running at equal speeds![/quote]

Everything looks normal (except that the XP 1600+ reports that SSE is not supported - no big deal). You are correct that your benchmarks look fine.

When you say they are factoring at equal speeds, I assume you mean ECM? Are they the same speed in both stage 1 and stage 2? Would an ECM benchmark from my P4 machine be helpful?

ET_ 2002-09-16 17:12

I have a question...

Why your XP1600 runs faster than XP1700?

Luigi :question:

paulunderwood 2002-09-16 18:02

I am running ECM on P32340 on both XP1600+ and P4; M32340 on XP1700+.

When I run the benchmark on the P4 it reports SSE2 but when I run ECM it is absent.

I have added CpuSupportsSSE=1 to XP1600+'s local.ini and it reports SSE in both the bench test and ECM.

I will try M32340 on the P4 and see it runs any quicker than the XP1700+.

So it seems I am getting dodgy readings when I am doing "P" ECM tests on the P4.

P4 timing for P32340 at B1=1M:
stage 1: 1772.832 sec
stage 2: 697.860 sec

P4 timing for M32340 at B1=1M (shows SSE2):
stage 1: 459.727 sec
stage 2: 198.783 sec

XP1700+ timing for M32340 at B1=M1:
stage 1: 1037.??? sec
stage 2: 479.126 sec

XP1600+ timing for P32340 at B1=1M:
stage 1: 1502.217 sec
stage 2: 688.680 sec

Lesson: run "M" ECM on P4 and "P" ECM on XPs fo max speed :-)

Question: Why does the P4 not use SSE2 for "P" ECM?


All times are UTC. The time now is 12:39.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.