The 2 chips were some years apart. Btw you have any pics of the cards and are you planning on testing more cards?
Tbf, TNT1 doesn't seem like a very useful retro card to me, but at least your had less of the weird artefacting you are decribing 😜

Funnily enough, MX400 had only 50% memory speed increase over original TNT. Even slightly worse, if we consider that TNT had SGRAM options and GeForce 2 MX don't.

Tetrium wrote on 2020-06-24, 14:36:

The 2 chips were some years apart. Btw you have any pics of the cards and are you planning on testing more cards?
Tbf, TNT1 doesn't seem like a very useful retro card to me, but at least your had less of the weird artefacting you are decribing 😜

I dunno I love my tnt1, goes great with my Pentium ii

Tetrium wrote on 2020-06-24, 14:36:

The 2 chips were some years apart. Btw you have any pics of the cards and are you planning on testing more cards?
Tbf, TNT1 doesn't seem like a very useful retro card to me, but at least your had less of the weird artefacting you are decribing 😜

I meant to mention that in my post - edited. Amazing the difference that even 2.5 years made when we were near "peak advancement rate" for graphics improvements.

Top card is the MX400, bottom is the TNT1. I had a spare 80mm fan blowing right at the heatsink of both cards for the tests. The MX400 looks like it might need re-capped; that could be the source of the artifacting?

I did this test on a whim; I'd add more into it if I have what people want to see tested. I've got some less common cards at the moment - NVS280 PCI, Quadro4 XGL, Voodoo3 2000 (common), SiS 305 AGP, Savage4 AGP, a leftover Radeon 7000 VE with a VESA P&D (Plug & Display) connector (I think AKA M1?).

candle_86 wrote on 2020-06-24, 15:13:

Tetrium wrote on 2020-06-24, 14:36:

The 2 chips were some years apart. Btw you have any pics of the cards and are you planning on testing more cards?
Tbf, TNT1 doesn't seem like a very useful retro card to me, but at least your had less of the weird artefacting you are decribing 😜

I dunno I love my tnt1, goes great with my Pentium ii

I never really much liked mine. It was also in a P2.
It worked, but 32-bit was basically useless due to how slow it would run.
I didn't hate the card either. It seemed to get old quite quickly like everything else in those days.

NautilusComputer wrote on 2020-06-24, 17:13:

I meant to mention that in my post - edited. Amazing the difference that even 2.5 years made when we were near "peak advancement […]
Show full quote

Tetrium wrote on 2020-06-24, 14:36:

The 2 chips were some years apart. Btw you have any pics of the cards and are you planning on testing more cards?
Tbf, TNT1 doesn't seem like a very useful retro card to me, but at least your had less of the weird artefacting you are decribing 😜

I meant to mention that in my post - edited. Amazing the difference that even 2.5 years made when we were near "peak advancement rate" for graphics improvements.

Top card is the MX400, bottom is the TNT1. I had a spare 80mm fan blowing right at the heatsink of both cards for the tests. The MX400 looks like it might need re-capped; that could be the source of the artifacting?

I did this test on a whim; I'd add more into it if I have what people want to see tested. I've got some less common cards at the moment - NVS280 PCI, Quadro4 XGL, Voodoo3 2000 (common), SiS 305 AGP, Savage4 AGP, a leftover Radeon 7000 VE with a VESA P&D (Plug & Display) connector (I think AKA M1?).

What I find interesting and funny is that, at first glance, these 2 cards seem quite similar superficially, but performance differed so much even just 2 1/2 years apart and one of the 2 is actually a budget card 🤣

I never had a SiS 305 I think, but SiS was certainly not the fastest 😜

It's quite possible that the caps were the reason for the artefacting.

Tetrium wrote on 2020-06-24, 20:46:

I never really much liked mine. It was also in a P2. It worked, but 32-bit was basically useless due to how slow it would run. I […]
Show full quote

candle_86 wrote on 2020-06-24, 15:13:

Tetrium wrote on 2020-06-24, 14:36:

The 2 chips were some years apart. Btw you have any pics of the cards and are you planning on testing more cards?
Tbf, TNT1 doesn't seem like a very useful retro card to me, but at least your had less of the weird artefacting you are decribing 😜

I dunno I love my tnt1, goes great with my Pentium ii

I never really much liked mine. It was also in a P2.
It worked, but 32-bit was basically useless due to how slow it would run.
I didn't hate the card either. It seemed to get old quite quickly like everything else in those days.

Loved mine at the time, also a P2. But I had just come from a 486 so any 3D was impressive. I distinctly remember the cop lights in NFS3.
But yeh didn't take long for it to start holding back rest of the system. Unless your into period builds I don't see a use for them either

GeForce 2 MX400 doubled TMUs per pixel pipe and doubled core clock speed. In scenarios, where memory bandwidth isn't limiting factor, it should be 4x faster.

The Serpent Rider wrote on 2020-06-25, 14:09:

GeForce 2 MX400 doubled TMUs per pixel pipe and doubled core clock speed. In scenarios, where memory bandwidth isn't limiting factor, it should be 4x faster.

Definitely not getting a full 4x; could be the low AGP support partially (board only 2x I think)? Could be bad caps or the Duron 1200 CPU.

I'm kind of glad caps didn't blow up during the benchmark.

Try with the oldest driver possible for TNT.

NautilusComputer wrote on 2020-06-25, 20:30:

The Serpent Rider wrote on 2020-06-25, 14:09:

GeForce 2 MX400 doubled TMUs per pixel pipe and doubled core clock speed. In scenarios, where memory bandwidth isn't limiting factor, it should be 4x faster.

Definitely not getting a full 4x; could be the low AGP support partially (board only 2x I think)? Could be bad caps or the Duron 1200 CPU.

No there is no way to escape the memory bandwidth problem of the geforce2 series, better way to see this is get an SDR mx 420 and click it at mx400 speeds. Only difference is the mx420 has the crossbar memory and hyperz.

This 3dmark Score is Really Low for a TNT1.
Seems you run too new driver. I can Even Play ut2003 on it...

And one of the cards is clearly defective.
Look at the capacitors.

Say what?:

His MX400 is 128bit based on 32 bit memory chips using 4 of them.

MX420 is 166MHz on the memory clock and if other guy has Geforce2 MX400 SDR with 128 bit bus, already running at 200MHz on SDR and MX400 core is also running at 200MHz. I was thinking about memory bandwidth.

To get bit more, should been looking at Geforce4 MX440 and bump both core and memory clocks little more and better cooling.

But I noticed the memory chips of MX400 is 4ns means could do around 250 to 300MHz?

Cheers,

pentiumspeed wrote on 2020-06-25, 22:41:

Say what?: […]
Show full quote

Say what?:

His MX400 is 128bit based on 32 bit memory chips using 4 of them.

MX420 is 166MHz on the memory clock and if other guy has Geforce2 MX400 SDR with 128 bit bus, already running at 200MHz on SDR and MX400 core is also running at 200MHz. I was thinking about memory bandwidth.

To get bit more, should been looking at Geforce4 MX440 and bump both core and memory clocks little more and better cooling.

But I noticed the memory chips of MX400 is 4ns means could do around 250 to 300MHz?

Cheers,

Mx420 is 128bit if made with SDR, while the 440 is always DDR. That's why I suggested it. The only big difference between nv15 and nv17 is the memory controller and an mx420 is an mx400 with a GeForce 4ti memory controller

Your Geforce2 MX400 has several bloated capacitors.

Geforce4 MX420 and MX440 are clocked differently and are same chips. MX440 is easier to get as well.

Cheers,

Proper TNT results (Pentium II 450):

pentiumspeed wrote on 2020-06-26, 01:00:

Your Geforce2 MX400 has several bloated capacitors.

Geforce4 MX420 and MX440 are clocked differently and are same chips. MX440 is easier to get as well.

Cheers,

Ok again mx 420 is an SDR card or 64bit ddr, mx 440 is a DDR card with either 128 or 64bit ddr. Because the mx420 is available in SDR you can compare it against the the mx 400 quite easily

NautilusComputer wrote on 2020-06-24, 17:13:

I meant to mention that in my post - edited. Amazing the difference that even 2.5 years made when we were near "peak advancement […]
Show full quote

Tetrium wrote on 2020-06-24, 14:36:

The 2 chips were some years apart. Btw you have any pics of the cards and are you planning on testing more cards?
Tbf, TNT1 doesn't seem like a very useful retro card to me, but at least your had less of the weird artefacting you are decribing 😜

I meant to mention that in my post - edited. Amazing the difference that even 2.5 years made when we were near "peak advancement rate" for graphics improvements.

Top card is the MX400, bottom is the TNT1. I had a spare 80mm fan blowing right at the heatsink of both cards for the tests. The MX400 looks like it might need re-capped; that could be the source of the artifacting?

I did this test on a whim; I'd add more into it if I have what people want to see tested. I've got some less common cards at the moment - NVS280 PCI, Quadro4 XGL, Voodoo3 2000 (common), SiS 305 AGP, Savage4 AGP, a leftover Radeon 7000 VE with a VESA P&D (Plug & Display) connector (I think AKA M1?).

I'd like to see the SiS 305 and the Savage 4, nice to visit some less usual cards.