XGI Volari
matrox g450
Savage 4

Warlord wrote on 2021-03-03, 06:12:

XGI Volari

verified working? need more details where do I add to list? 9x support? which model?

never said verified working. Doubt you will find a XGI Volari V3XT PCI anyways

Warlord wrote on 2021-03-03, 06:19:

never said verified working. Doubt you will find a XGI Volari V3XT PCI anyways

Okay, I didn’t know what category to put it in. I’ll put it in the untested category

There is a xgi volari on ebay right now 16mb pn pcivgav7 made by startech

New chinese fx5500 not boot on my 430tx (KM-T5-T1 REV:1.32).

Chinese FX5500 don't use buffer chips, which drops compatibility with old motherboards even further.

Many years ago my first K6-2+ build with an Asus P55T2P4 build used a Pine TNT2 32 meg PCI card, add it to the list. I've also used a GeForce 4 MX420 64 meg card with a TXP4 successfully.

Radeon 64mb 7000 pci seems to work without issues on windows ME and windows 7 using the catalyst 6.2 driver

Sphere478 wrote on 2021-03-03, 04:37:

here is a list of cards with reported compatibility problems bad enough to be considered not working: […]
Show full quote

here is a list of cards with reported compatibility problems bad enough to be considered not working:

-rage 128 no display on boot (2 people report)
-cirris logic cl-gd5430 no display on boot
-trident tgui9680-1 no display on boot

It's really strange. All these cards are from that period and should work flawlessly with 430tx at least.

For me, period correct seems to be the best shout because anything newer is really bottlenecked. Pentium II onwards is where CPU and GPU performance really skyrocketed.

So in my machines, I've cycled around the following cards:

- Matrox G200
- ATi RAGE PRO Turbo (Mine is an All-in-Wonder variant)
- S3 ViRGE DX (Had a few no-name cards, plus a Diamond Stealth 3D 2000)

All paired with and without a Voodoo 1 card.

RichB93 wrote on 2021-03-06, 14:05:

For me, period correct seems to be the best shout because anything newer is really bottlenecked. Pentium II onwards is where CP […]
Show full quote

For me, period correct seems to be the best shout because anything newer is really bottlenecked. Pentium II onwards is where CPU and GPU performance really skyrocketed.

So in my machines, I've cycled around the following cards:

- Matrox G200
- ATi RAGE PRO Turbo (Mine is an All-in-Wonder variant)
- S3 ViRGE DX (Had a few no-name cards, plus a Diamond Stealth 3D 2000)

All paired with and without a Voodoo 1 card.

It’s never that simple though as calling something a bottleneck and the term is misleading. There are way too many aspects to compare it to liquid moving in a circuit.

There is driver optimization, processing that happens on card without much need for througput.

In general a faster card if you can get the drivers to work will usually give higher benchmarks. I mean they released pci cards all the way up to the gt610 and people were crying about bottlenecks back in the radeon 7000 days yet the 610 wipes the floor with the 7500 on pci

But sometimes you’ll see some game benchmarks saying the opposite where a older card and drivers are showing higher performance because that game was made to better use that old combination. Though usually once a card becomes a lot faster than that old hardware if you can get the game to work it will still beat the optimized hardware.

I avoid over simplify things, and calling things bottlenecks it’s much more complicated than that and does a disservice to understanding the issue.

What should I try next? I’m leaning toward a radeon x series on adapter..

Radeons seem to have good luck. Though I miss the wide screen bios on the 6200 the 7000 and x1300 show 4:3 ratio

Sphere478 wrote on 2021-03-06, 16:38:

It’s never that simple though as calling something a bottleneck and the term is misleading. There are way too many aspects to co […]
Show full quote

RichB93 wrote on 2021-03-06, 14:05:

For me, period correct seems to be the best shout because anything newer is really bottlenecked. Pentium II onwards is where CP […]
Show full quote

For me, period correct seems to be the best shout because anything newer is really bottlenecked. Pentium II onwards is where CPU and GPU performance really skyrocketed.

So in my machines, I've cycled around the following cards:

- Matrox G200
- ATi RAGE PRO Turbo (Mine is an All-in-Wonder variant)
- S3 ViRGE DX (Had a few no-name cards, plus a Diamond Stealth 3D 2000)

All paired with and without a Voodoo 1 card.

It’s never that simple though as calling something a bottleneck and the term is misleading. There are way too many aspects to compare it to liquid moving in a circuit.

There is driver optimization, processing that happens on card without much need for througput.

In general a faster card if you can get the drivers to work will usually give higher benchmarks. I mean they released pci cards all the way up to the gt610 and people were crying about bottlenecks back in the radeon 7000 days yet the 610 wipes the floor with the 7500 on pci

But sometimes you’ll see some game benchmarks saying the opposite where a older card and drivers are showing higher performance because that game was made to better use that old combination. Though usually once a card becomes a lot faster than that old hardware if you can get the game to work it will still beat the optimized hardware.

I avoid over simplify things, and calling things bottlenecks it’s much more complicated than that and does a disservice to understanding the issue.

I look forward to seeing some game benchmarks with the massively newer cards you've mentioned. They literally are bottlenecked by a S7 machine. They have the potential to be faster than contemporaries of the time, but that doesn't mean that they are not bottlenecked by the system itself. Usually the increased complexities of their drivers create disadvantages too.

As far as analogies go, I think it's a perfectly apt one and is commonly used in computer science; if the system can't feed the card data fast enough, the system is bottlenecking it.

Sphere478 wrote on 2021-03-06, 16:38:

In general a faster card if you can get the drivers to work will usually give higher benchmarks. I mean they released pci cards all the way up to the gt610 and people were crying about bottlenecks back in the radeon 7000 days yet the 610 wipes the floor with the 7500 on pci

You are really mixing things up. A more powerful video card will wipe the floor with a less powerful one when CPU bottleneck isn't an issue. On a 430TX, a Voodoo 2/Voodoo 3 will wipe the floor with any other card (including the ones you mentioned). Why? The answer is the older drivers (which are optimized for slow CPUs) and Glide (a much more efficient and lightweight API).
Not to mention that the whole idea of using a Pentium MMX & Co is having as much compatibility as possible. And this is another area where older cards shine (particularly Voodoo cards). I myself am not a fan of the Voodoo 1 (I think that it's only decent on very early Pentiums), but Voodoo 2 or 3 are great for such a system (Voodoo 2 for better compatibility with early Glide games / Voodoo 3 for better performance, image quality and ability to play at higher resolutions).

RichB93 wrote on 2021-03-07, 12:03:

Sphere478 wrote on 2021-03-06, 16:38:

It’s never that simple though as calling something a bottleneck and the term is misleading. There are way too many aspects to co […]
Show full quote

RichB93 wrote on 2021-03-06, 14:05:

For me, period correct seems to be the best shout because anything newer is really bottlenecked. Pentium II onwards is where CP […]
Show full quote

For me, period correct seems to be the best shout because anything newer is really bottlenecked. Pentium II onwards is where CPU and GPU performance really skyrocketed.

So in my machines, I've cycled around the following cards:

- Matrox G200
- ATi RAGE PRO Turbo (Mine is an All-in-Wonder variant)
- S3 ViRGE DX (Had a few no-name cards, plus a Diamond Stealth 3D 2000)

All paired with and without a Voodoo 1 card.

It’s never that simple though as calling something a bottleneck and the term is misleading. There are way too many aspects to compare it to liquid moving in a circuit.

There is driver optimization, processing that happens on card without much need for througput.

In general a faster card if you can get the drivers to work will usually give higher benchmarks. I mean they released pci cards all the way up to the gt610 and people were crying about bottlenecks back in the radeon 7000 days yet the 610 wipes the floor with the 7500 on pci

But sometimes you’ll see some game benchmarks saying the opposite where a older card and drivers are showing higher performance because that game was made to better use that old combination. Though usually once a card becomes a lot faster than that old hardware if you can get the game to work it will still beat the optimized hardware.

I avoid over simplify things, and calling things bottlenecks it’s much more complicated than that and does a disservice to understanding the issue.

I look forward to seeing some game benchmarks with the massively newer cards you've mentioned. They literally are bottlenecked by a S7 machine. They have the potential to be faster than contemporaries of the time, but that doesn't mean that they are not bottlenecked by the system itself. Usually the increased complexities of their drivers create disadvantages too.

As far as analogies go, I think it's a perfectly apt one and is commonly used in computer science; if the system can't feed the card data fast enough, the system is bottlenecking it.

So far the only card I’ve gotten to actually work on the k6 3+ 430tx setup is the radeon 7000 so a comprehensive comparison with the growing selection of cards I have just isn’t something I’m going to be able to provide. The 6200 seems to work alright in windows 7 but windows 7 is taking up a decent amount of resources so on this system I’m using 7 for windows related tasks, restoring ME when the drivers crash the install, etc 🤣. And windows ME is where I am trying to set up the games I’m playing on this computer.

It’s not so black and white that you can tell me a specific card where there will be no performance increase or decrease there are other factors. Like the other post mentions about the voodoo cards. Driver and chip optimizations can sometimes make a less spec card perform better than a higher spec card. It’s more complicated than the oversimplified misleading term bottleneck

If it was that simple then you would be able to come up with a list of cards rated by processing power and you would get a increase in a specific output until you got to a specific point and there would be no more and you could tell me what card that would be and all games and benchmarks would rely on that one criteria But real world is: there is this game does better up until this point, but that game does better with this card but we get better performance with this card at these resolutions unles we use this driver yet this one really old card does better with this driver and not that driver however this game likes this really new card but our mining program likes this card the most but this game likes this old card and driver however this benchmark likes this card the best but this benchmark likes this card and driver the best, etc

It’s a oversimplification. Especially with video cards. There is too much going on on a video card to reduce it to a single value and it’s a disservice to accuracy to do so in my opinion. The voodoo comment I feel shows this wonderfully how there is more going on here than something that can be simplified that much.

A better application for the term is networking, storage media etc, but even that is still a oversimplification. There are different protocols that can increase throuput, more than one component comes into play and sometimes you’ll get a performance boost from upgrading any one of several parts which dissagrees with the notion of one item being the complete cause of the slow down and there being a marked point where the next item is the issue like saying well, this network card is the issue and upgrading the cpu won’t do anything but the truth is upgrading either one shows a improvement. So the fact was that they both helped. I just have a issue with the oversimplification of the term. It breeds a shallow level of understanding of the broader issues at play.

Sphere478 wrote on 2021-03-07, 13:27:

It’s more complicated than the oversimplified misleading term bottleneck

There is nothing misleading about the term bottleneck.

If you pair a high end graphics card with a low end CPU, you will be CPU bottlenecked in most resolutions. This can be easily verified by running a CPU intensive game (e.g. Unreal Tournament '99) at 640x480, 800x600, 1024x768 and 1600x1200. If you compare the results, it's very likely that you will get the exact same frame rate in the first three resolutions, within margin of error. Only at the highest resolution will the graphics card start hitting its limit.

Joseph_Joestar wrote on 2021-03-07, 14:12:

Sphere478 wrote on 2021-03-07, 13:27:

It’s more complicated than the oversimplified misleading term bottleneck

There is nothing misleading about the term bottleneck.

If you pair a high end graphics card with a low end CPU, you will be CPU bottlenecked in most resolutions. This can be easily verified by running a CPU intensive game (e.g. Unreal Tournament '99) at 640x480, 800x600, 1024x768 and 1600x1200. If you compare the results, it's very likely that you will get the exact same frame rate in the first three resolutions, within margin of error. Only at the highest resolution will the graphics card start hitting its limit.

You would need to use very specific examples for it to be totally acurate.

The statement for example: a gtx 1060 is bottlenecked by a core two duo

Okay...

The mining program doesn’t really care that much..

But 3dmark 06 might care a little bit more. But still get a even better score with a 2070

It’s a term used way too often without the relevant specifics it fosters a limited understanding of a more complex situation when many people use it. So often the way people are using it makes it seem like everything in your system has a number 1-10 and you have to make everything a 7 or a 9 and all programs and tasks will increase linerally with that and unless you make them all a 8 then nothing will get any faster if you still have a 7 in there somewhere. It’s oversimplification when people use the term without a very specific and detailed example and explanation of the underlying factors at play.

Sphere478, at this point you are basically arguing semantics 😀

First of all, your comparison with the Core 2 Duo / GTX 1060 / 3DMark06 is not really relevant for the 430TX discussion, because the performance jump from a Core 2 Duo to a modern CPU, although it might seem extreme at first, it's really not.

I mean, sure, we now have 12/16 cores or more, but the performance of each of those modern cores is probably just 3 or 4 times better than the performance of one of the cores in the Core 2 Duo E8600. This really isn't a lot when taking into account the fact that 12 years have passed since that CPU was released. Furthermore, the Core 2 Duo CPU has a lot of the basic instructions that are in use even today and for which drivers are still optimized (SSE, SSE2, SSE3, SSSE3, SSE4.1, etc). Since 3DMark06 is mostly a single threaded benchmark (from what I remember it strugles to even use two threads), the difference between a modern CPU and an old Core 2 Duo will probably not be that extreme (which is why 3DMark06 should not be used for modern benchmarking, a warning that Futuremark themselves have on their website).

Now let's compare the Pentium MMX to a Pentium 4 Northwood 3 GHz. These CPUs were released just 5 or 6 years apart, but the Pentium 4... is probably 2 orders of magnitude faster than the MMX (maybe much more in certain workloads). Not to mention the lack of SSE/SSE2 instructions on the Pentium MMX. What if we now compare it to a Core 2 Duo (8 years apart)? 😀

There's no other way of putting it: a Pentium MMX will be a SEVERE bottleneck for newer video cards (when talking about games, because we're not talking about Microsoft Word, are we?).
Generally speaking, the Pentium MMX bottlenecks most video cards 😀, but newer video cards (2001+) will increase this handicap even further, since their drivers expect CPUs with newer instructions (SSE/SSE2) and substantially more raw power.

bloodem wrote on 2021-03-07, 14:59:

There's no other way of putting it: a Pentium MMX will be a SEVERE bottleneck for newer video cards (when talking about games, because we're not talking about Microsoft Word, are we?).
Generally speaking, the Pentium MMX bottlenecks most video cards 😀, but newer video cards (2001+) will increase this handicap even further, since their drivers expect CPUs with newer instructions (SSE/SSE2) and substantially more raw power.

I agree, most machines based on P5 won't be able to keep up with 'newer' video cards (newer meaning cards that came out in the later 90s). There are tons of threads about this on vogons but personally testing a tnt2 yielded basically the same framerate in opengl stars test regardless of the resolution i chose, which seems to me like a bottleneck outside of the graphics card.

Anyway I'm a bit confused as to the thread's topic - is it to just discuss which cards *can* POST and "work," or is it a discussion about their performance and/or viability in a socket 7 system? If the former - I think it should be noted that even if the chipset is the same, certain motherboards may just not work with specific cards. I have a socket 7 which refuses to work with the aforementioned PCI TNT2 for example, but works fine with another system based on the same chipset.