Optomized S5 SSE3

Sven Ullrich
Sven Ullrich
Joined: 20 Feb 05
Posts: 3
Credit: 1252647
RAC: 0

Now seven valid S5T0003

Message 39027 in response to message 38972

Now seven valid S5T0003 results on my Pentium Mobile 740.
You can see it here results .
speed-up 1-2%

LiborA
LiborA
Joined: 8 Dec 05
Posts: 74
Credit: 337135
RAC: 0

RE: Thanks for all... But

Message 39028 in response to message 39026

Quote:

Thanks for all...

But could anybody do a time measurement with a same (short) wu?

I would like to know a reference time for the official app and i would like to know a time for S5T0304 (it seems to be valid) or S5T0305 (i will upload soon).

I expect they have bigger speed improvement than 10%...

Yes, I try 4 WUs, wait a 10 minute (the 3th WU will finish), but I have only SSE2 CPU and therefore i'm testing only S5T0003 not S5T030X.

Raijin1979
Raijin1979
Joined: 25 Aug 05
Posts: 1
Credit: 12759
RAC: 0

that's my Time standard

that's my Time

standard app. ca.8h
Result with Standard App.

opt. app.(S5T0303) ca. 6.5h
Result with Opti. App
my cpu is a A64 3000@3800+ with SSE3

Akos Fekete
Akos Fekete
Joined: 13 Nov 05
Posts: 561
Credit: 4527270
RAC: 0

S5T0305.dat - eliminated

Message 39030 in response to message 39009

S5T0305.dat

- eliminated double jumps
- reduced amount of FPU macro ops
- removed double loads on general purpose registers

- better SSE register usage
- reduced memory and integer register usage

CPU: SSE compatible

Sven Ullrich
Sven Ullrich
Joined: 20 Feb 05
Posts: 3
Credit: 1252647
RAC: 0

RE: S5T0305.dat -

Message 39031 in response to message 39030

Quote:

S5T0305.dat

- eliminated double jumps
- reduced amount of FPU macro ops
- removed double loads on general purpose registers

- better SSE register usage
- reduced memory and integer register usage

CPU: SSE compatible


Run S5T0305 on SSE2 cpus?

LiborA
LiborA
Joined: 8 Dec 05
Posts: 74
Credit: 337135
RAC: 0

RE: RE: Thanks for

Message 39032 in response to message 39028

Quote:
Quote:

Thanks for all...

But could anybody do a time measurement with a same (short) wu?

I would like to know a reference time for the official app and i would like to know a time for S5T0304 (it seems to be valid) or S5T0305 (i will upload soon).

I expect they have bigger speed improvement than 10%...

Yes, I try 4 WUs, wait a 10 minute (the 3th WU will finish), but I have only SSE2 CPU and therefore i'm testing only S5T0003 not S5T030X.

I open new thread fot this here

LiborA
LiborA
Joined: 8 Dec 05
Posts: 74
Credit: 337135
RAC: 0

RE: RE: S5T0305.dat -

Message 39033 in response to message 39031

Quote:
Quote:

S5T0305.dat

- eliminated double jumps
- reduced amount of FPU macro ops
- removed double loads on general purpose registers

- better SSE register usage
- reduced memory and integer register usage

CPU: SSE compatible


Run S5T0305 on SSE2 cpus?

NO!!!

EDIT: Sorry, I'm not sure yet (after reading this post)

_heinz
_heinz
Joined: 4 Jan 06
Posts: 79
Credit: 130476
RAC: 0

Pentium4 2.66MHz

Pentium4 2.66MHz WinXP
S5S0003 long h1 Wu 12h 57min 55sec =46675 sec
Validate state Initial
Claimed credit 163.234953703704
34674064
will now try S5T0305 with a long h1 Wu

Akos Fekete
Akos Fekete
Joined: 13 Nov 05
Posts: 561
Credit: 4527270
RAC: 0

RE: RE: S5T0305.dat CPU:

Message 39035 in response to message 39031

Quote:
Quote:

S5T0305.dat

CPU: SSE compatible

Run S5T0305 on SSE2 cpus?

The SSE2 CPUs can execute also the SSE instructions, so you can run it on all SSE2 CPUs without compatibility problems.

Akos Fekete
Akos Fekete
Joined: 13 Nov 05
Posts: 561
Credit: 4527270
RAC: 0

S5T0306.dat - eliminated

Message 39036 in response to message 39030

S5T0306.dat

- eliminated double jumps
- reduced amount of FPU macro ops
- removed double loads on general purpose registers

- better SSE register usage
- reduced memory and integer register usage
- optimized branch structure

CPU: SSE compatible

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.