supercordo wrote on 2023-11-17, 20:27:

Benchmarking is fun!!!

It definately is! I'm reliving my early 20's again 🤣 Except back then it was a Northwood 1.6 @ 2.4 😁

supercordo wrote on 2023-11-17, 20:27:

Benchmarking is fun!!!

Is that your Ti4600 on the picture ? How does it perform compared to the Ti500 ?

acl wrote on 2023-11-17, 21:06:

supercordo wrote on 2023-11-17, 20:27:

Benchmarking is fun!!!

Is that your Ti4600 on the picture ? How does it perform compared to the Ti500 ?

Its mine! The Ti 500 is faster since it can use older drivers. All the drivers for the Ti 4600 have more cpu overhead. I havnt tested it yet on my S423 tho.

supercordo wrote on 2023-11-17, 21:15:

acl wrote on 2023-11-17, 21:06:

supercordo wrote on 2023-11-17, 20:27:

Benchmarking is fun!!!

Is that your Ti4600 on the picture ? How does it perform compared to the Ti500 ?

Its mine! The Ti 500 is faster since it can use older drivers. All the drivers for the Ti 4600 have more cpu overhead. I havnt tested it yet on my S423 tho.

I was also thinking about this sentence from wikipedia :

It is believed that the fixed-function T&L hardware from GeForce 2 was still included on the [GeForce 3] chip for use with Direct3D 7.0 applications, as the single vertex shader was not fast enough to emulate it yet

This might give an advantage for GeForce3 over the 4 on pure DX7 workloads. The DX7 rendering over programmable pipeline might add some CPU overhead with more complex drivers on the GeForce4.
But not sure if it's true or just a myth

OK, now ... 118 FSB, 1.875V according to BIOS (I don't like what CPU-Z reads). Memory timings tidied up, AGP @ 72 MHz ...

Where to go for the last 307 points?

H3nrik V! wrote on 2023-11-17, 21:43:

OK, now ... 118 FSB, 1.875V according to BIOS (I don't like what CPU-Z reads). Memory timings tidied up, AGP @ 72 MHz ...

Where to go for the last 307 points?

Updated! You could start looking at optimizing windows?

H3nrik V! wrote on 2023-11-17, 21:43:

OK, now ... 118 FSB, 1.875V according to BIOS (I don't like what CPU-Z reads). Memory timings tidied up, AGP @ 72 MHz ...

Where to go for the last 307 points?

Drivers options can help too : The "quality vs speed" slider. MipMap quality, etc... just set everything to "fast" 😁

supercordo wrote on 2023-11-17, 21:56:

H3nrik V! wrote on 2023-11-17, 21:43:

OK, now ... 118 FSB, 1.875V according to BIOS (I don't like what CPU-Z reads). Memory timings tidied up, AGP @ 72 MHz ...

Where to go for the last 307 points?

Updated! You could start looking at optimizing windows?

That's probably outside my capabilites 🤣

acl wrote on 2023-11-17, 22:02:

H3nrik V! wrote on 2023-11-17, 21:43:

OK, now ... 118 FSB, 1.875V according to BIOS (I don't like what CPU-Z reads). Memory timings tidied up, AGP @ 72 MHz ...

Where to go for the last 307 points?

Drivers options can help too : The "quality vs speed" slider. MipMap quality, etc... just set everything to "fast" 😁

Smells almost like cheating, buuuut, the rules doesn't say that I can't 😎

The first post has been updated with the link to the post for each score. I think its awesome.

Yay! Looks like a top-3 in 3DMark 2000 - at least because not many has run it 🤣

H3nrik V! wrote on 2023-11-18, 09:45:

Yay! Looks like a top-3 in 3DMark 2000 - at least because not many has run it 🤣

Score updated! Nice.

Its on now!!! Just need my board to get here.

If you dont have a test bench yet! DimasTech went out of buisness so these are hard to find. Seller says they have 7 available in New York. Mine arrives on Monday, Ill let you know if i like it. https://www.ebay.com/itm/145010031026

A 1.5 Willamette arrived today. Was able to increase the FSB. Was only able to beat my old score by 4. I'm sure ill be able to push it further with the P4P800. It seems 3dMark2000 likes more MHz vs FSB.

supercordo wrote on 2023-11-19, 05:06:

If you dont have a test bench yet! DimasTech went out of buisness so these are hard to find. Seller says they have 7 available in New York. Mine arrives on Monday, Ill let you know if i like it. https://www.ebay.com/itm/145010031026

In the name of all that is holy, THAT is a piece of beauty! It's a shame I'm in the EU, though ...

FSB yanked up to 120, CPU @2160. Tight memory timings, GPU up to 305 MHz, driver setting to fastest. Target 10000 achieved. Funny enough, 3DM2000 responds bad to these settings, giving in the 8500-ballpark ...

What next - try another speed P4 stepping D0 or try to see what an E0 Celery will clock to?

supercordo wrote on 2023-11-19, 07:56:

A 1.5 Willamette arrived today. Was able to increase the FSB. Was only able to beat my old score by 4. I'm sure ill be able to push it further with the P4P800. It seems 3dMark2000 likes more MHz vs FSB.

What board are you running here?

making a lateral consideration, can the P4 be seen as the P60? Not in terms of architecture because the Pentium continued until 233 but in terms of the uniqueness of the socket/platform? The P4 wasn't there before and will never be there again after its disappearance, right?

AlessandroB wrote on 2023-11-19, 11:44:

making a lateral consideration, can the P4 be seen as the P60? Not in terms of architecture because the Pentium continued until 233 but in terms of the uniqueness of the socket/platform? The P4 wasn't there before and will never be there again after its disappearance, right?

Desktop Pentium 4 was available on 3 sockets, 423,478,775 .
Probably the most special one is 423 with its small range of processors released and its very short lifetime.
There is a version of socket 423 boards with sd ram instead of RD ram. Thats probably the most exotic version... not good, but exotic.

Or are you speaking about the "long pipeline" thing?