Was 3DMark99Max a lot more popular than the Final Reality v1 benchmark?

well, I've never heard of the 2nd one until this, so.... yes? The later 3dmarks are the defacto standard for system/video card benchmarking. so I would include it.

I used both at that time. 3Dmark survived, so it must have been more known.

Based on the feedback I received, I will include extra charts for Quake 1, Quake 2 - Software, Quake 2 - OpenGL, DOOM, 3DMark99Max, and PassMark MMX. These charts, as well as averaged ALU, FPU, Overall charts have now been completed. I need to check the data for errors before starting a new thread, but on first glance it all looks good.

The charts should be able to display directly into the web browser. I just fixed the Ultimate 486 Benchmark Comparison thread so that the charts display directly now. I'm glad this is just about over.

Reading from your first post, did you ever find someone with a Socket 4 system? I've got a working system with a P60 (SX835 /w FDIV bug). I can also set the jumper to P66 which works. I don't have a G200 PCI card though.

I also have the Slot 1 Pentium II 450mhz

vetz wrote:

Reading from your first post, did you ever find someone with a Socket 4 system? I've got a working system with a P60 (SX835 /w FDIV bug). I can also set the jumper to P66 which works. I don't have a G200 PCI card though.

I also have the Slot 1 Pentium II 450mhz

You'd need G200 PCI card, preferablly with 16 MB. I did not find someone to run the P60/66 or PII-450. I synthesised the P60/66 data for a socket 7 environment, so it is probably 5-10% faster than it would be on a socket 4. If you can get results for the P60/66 and PII-450, I will add them. Thank you for your interest!

That is going to be a problem. Any preferance on other PCI cards? I guess the Millennium II would be a good fit on the socket 4 system (since the most important results would be synthetic and Quake software performance anyway).

For the PII I could go with the G200 AGP, but here I think it is only 8 MB.

I thought I had a p2-450... Let me double check. Also, I'm willing to lend vetz my g200 if he promises to mail it back to me when he is done.

Vetz - Yes, we need a PCI Matrox G200. All 177 CPUs have been tested using the same hardware. The exception with the hardware is that a different motherboard was needed when testing different socket CPUs, e.g. socket 7, socket 8, slot 1, etc. The motherboards employed thus far have been,

1 Biostar MB-8433UUD (UMC 8881F/8886BF) - socket 3
2 Biostar MB-8500TTD (Intel 430TX) - socket 7
3 FIC PA-2013 (VIA Apollo MVP3 - 598AT / 586B) - super socket 7
4 Asus P3V4X (VIA Apollo Pro 133A - 694X / 596B) - slot 1
5 Asus TUSL2-C (Intel 815EP) - socket 370
6 Gigabyte 6VX7-4X (VIA Apollo Pro 133A - 694X / 686A) - socket 370
7 MSI MS-6167 (AMD-750 Irongate - 751 / 756) - slot A
8 Asus P/I-P65UP5 with C-P6ND CPU Card (Intel 440FX) - socket 8
9 ECS P5GX-M (MediaGX + CX5530) - socket 7 GXM
10 Xeon motherboard - slot 2

I had to use two socket 370 boards only becuase the VIA Nehemiah CPUs only worked well on a particular motherboard, whereby the Samuel-Ezra cpus also only worked well on a particular motherboard. In no instance did any one of my socket 370 motherboards work with the full spectrum of VIA socket 370 CPUs.

Luckybob - could you fill in the blanks for the Xeon motherboard you used?

I thought I had finished up the charts, but forgot about the 133 MHz Challenge charts. I guess I have more work to do.

Also, for 177 CPUs on a single chart, it requires a lot of scrolling and can be somewhat difficult to get a good sense of the comparison. I wonder if there is a better way to represent this data, like with a single sheet of text, or similar. Open to suggestions. I've included a sample of this dilemma.

in stead of all 177 on one image, maybe cherry pick the "transitions" between sockets. Or separate them on the basis of mhz. So the 200's & 233's are on one page. 233-333 on page 2, etc

The transition by freq. can be tough because it breaks the continuity of scores. A P55C at 300 MHz has the performance of a Cyrix MII 233 MHz in Integer performance, but it is the flip case for floating point, whereby a Cyrix MII 233 Mhz is about a P55C 166 MHz. I'll sleep on this for a bit. Luckily, there are only 20 CPUs in the 133 MHz challenge, so the chart can fit on one page easily. Those charts are now finished.

Here's a bit of a teaser - between the following CPU's at 133 MHz, which CPU do you think came in second to the Pentium in Quake 2? All 5 Pentiums are on top, but which non-Pentium brand CPU came in with the next highest score? I was surprised by this one.

IBM 5x86c, 133 MHz, 66 / 2.0x
Cyrix MediaGX, 133 MHz, 33 / 4.0x
IBM 6x86, 133 MHz, 66 / 2.0x
Cyrix 6x86MX, 133 MHz, 66 / 2.0x
AMD X5, 133 MHz, 33 / 4.0x
AMD K5, 133 Mhz, 66 / 2.0x
AMD K6, 133 MHz, 66 / 2.0x
AMD K6-2, 133 MHz, 66 / 2.0x
AMD K6-2+, 133 MHz, 66 / 2.0x
AMD K6-3+, 133 MHz, 66 / 2.0x
Intel DX4, 133 MHz, 66 / 2.0x
Intel Pentium P54C, 133 MHz, 66 / 2.0x
Intel Pentium P55C, 133 MHz, 66 / 2.0x
Intel Pentium Pro P6, 133 MHz / 256kb, 66 / 2.0x
Intel Pentium Pro P6, 133 MHz / 1mb, 66 / 2.0x
Intel PII, 133 MHz, 66 / 2.0x
IDT Winchip C6, 133 MHz, 66 / 2.0x
IDT Winchip2, 133 MHz, 50 / 2.66x
IDT Winchip2, 133 MHz, 66 / 2.0x
Rise mP6, 133 MHz, 66 / 2.0x

i might be biased, but i'm going to say p-pro 1mb. ^.^

Would it bee too much to make a separate chart for each 33 mhz?

luckybob wrote:

i might be biased, but i'm going to say p-pro 1mb. ^.^

Would it bee too much to make a separate chart for each 33 mhz?

The P-Pro is an Intel Pentium. As noted, all the Pentiums came out on top. The question is, which non-Pentium was next in line for the highest score? Anyone else care to guess?

I don't follow the second question. You want every CPU from 60-90 MHz on one page, then each CPU from 100 - 133 Mhz on one page, then each CPU from 150-180 MHz on a single page, etc? It is more of a clock-for-clock type of comparison like that. The 133 MHz challenge should cover all these possibilities. If I have understood you correctly, then perhaps every 66 or 100 MHz would be more suitable. If every 66 MHz, it would be 9 pages.

I don't really like the idea of breaking the ranking order of the CPUs, but perhaps we could break it into two parts, 60-300 MHz and 300-600 MHz? I was just going to make a seperate page every 20 CPUs and keep the CPUs ranked in ascending order. Maybe I can put two 20-CPU charts side-by-side on a single page. This would only need 5 pages (or 9 pgs if I only put 1 chart per page).

Attached is a sample of the massive amount of data collected for this project. I still come across an error every now and again, which takes a long time to fix now that the charts are plotted.

EDIT: Sample_Data.pdf file removed in favour of the completed Ultimate 686 Benchmark Comparison thread.

Any chance you could release it as a web document or Excel document where it is possible to select and sort on which data we want to see?
I see this as a better way to present it than everything into one PDF file, but ofc you can do both.

If my testing will require a G200 PCI I think I will have to pass. Luckybob, I guess you are based in the US? Shipping costs will not make it worthwhile since I live in Europe. We're talking 10-13 dollars for you one way and the same for me to return it.

vetz wrote:

Any chance you could release it as a web document or Excel document where it is possible to select and sort on which data we want to see? I see this as a better way to present it.

The PDF attachment is not really meant for data viewing, although if you were really persistant, I suppose you could extract some results. The PDF was only to give everyone an idea for the massiveness amount of data collected. The data is not yet ready to be presented in a viewable way. A new thread will be started when the results are made public. If you view the Ultimate 486 Benchmark Comparison, there is an all inclusive PDF article contained there-in which can be printed and all the data viewed easily on two sheets of paper. The purpose of this thread was to give others the opportunity to assist in this undertaking, particularly in areas whereby outsourced tested are required to benchmark missing CPUs.

My next steps are to compile the data into printable PDF pages, along with all 18 bar charts. I haven't decided if I'll add the monologue to the article this time, like I did with the Ultimate 486 Benchmark Comparison. I think the most interesting monologue will be the clock-for-clock comparisons at 133 MHz because you can infer the outcome of higher frequency combinations.

Vetz: If you are in Europe, I recall buying a G200 from the UK for 4.5 pounds not too long ago. Otherwise, we are still waiting on Janoz to get some results over to us. I think he can only test the P60 because he has been unable to find an correct motherboard oscillator for use with the P66. Janoz does have a PCI Matrox G200 w/16 MB of RAM. He is in the Netherlands. Maybe you two can combine resources and get these P5 chips tested?

Regarding Quake 2: I suppose you mean software rendering or OpenGL without 3DNow?
Then I'd say AMD K5.
Reasoning: The chip was quite fast per clock (faster then K6), it just didn't scale. And with a multiplier of 2.0 the on-die L2-cache of K6-2+/3+ is not as important. This is confirmed by the 133 MHz Challenge Swaaye started.

No chance for X5 or DX4. Cyrix and IBM have weak FPUs. Winchip(2) is not even superscalar.

The only other contender would be the Rise mP6 which I can't find a lot of data for. Now this would be a surprise. The mP6 was limited by its small cache which might not have a big influence at 2*66 MHz.

idspispopd wrote:

Regarding Quake 2: I suppose you mean software rendering or OpenGL without 3DNow?
Then I'd say AMD K5.

You nailed it. The AMD K5, followed by the Rise mP6.
Yes, software rendering.

idspispopd wrote:

Regarding Quake 2: I suppose you mean software rendering or OpenGL without 3DNow? Then I'd say AMD K5. Reasoning: The chip was q […]
Show full quote

Regarding Quake 2: I suppose you mean software rendering or OpenGL without 3DNow?
Then I'd say AMD K5.
Reasoning: The chip was quite fast per clock (faster then K6), it just didn't scale. And with a multiplier of 2.0 the on-die L2-cache of K6-2+/3+ is not as important. This is confirmed by the 133 MHz Challenge Swaaye started.

No chance for X5 or DX4. Cyrix and IBM have weak FPUs. Winchip(2) is not even superscalar.

The only other contender would be the Rise mP6 which I can't find a lot of data for. Now this would be a surprise. The mP6 was limited by its small cache which might not have a big influence at 2*66 MHz.

From the few tests I have seen, the Rise MP6 performs worse on a per clock basis than the Winchip2. It did manage to reach higher speeds but Winchip2 had a 3DNow! unit that pushed it's gaming performance up in games that utilized it. The MP6 was a huge disappointment mostly because of it's tiny cache but also partly because it was rated relative to the Pentium MMX while it's competitors rated themselves relative to the Pentium II.

That makes sense. The Winchip2 performed much better than the mP6 in 3DMark99Max, which I assume utilises 3DNow!. I guess Quake 2 doesn't make use of 3DNow!?

feipoa wrote:

That makes sense. The Winchip2 performed much better than the mP6 in 3DMark99Max, which I assume utilises 3DNow!. I guess Quake 2 doesn't make use of 3DNow!?

I think you need to patch in 3DNow! optimizations for Q2.

Here's the press release from AMD.

http://www.amd.com/us/press-releases/Pages/Pr … elease_911.aspx

Try to find the 3DNow! patch then test again. The Winchip2 will probably beat it. The only reason the MP6 does so much better without it is because it actually has 3 MMX units in it so it can execute 3 MMX instructions per cycle.