VOGONS


The Ultimate 486 Benchmark Comparison

Topic actions

Reply 40 of 280, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Tetrium, nice find!

One model (ESA-TF486, MB103) has:
PS/2, 512 KB cache, 128 MB EDO RAM, 128 GB IDE harddrive support, Cyrix/AMD 5x86 support, IrDA, ALi chipset, AWARD BIOS, but only 1 PCI slot

While some (ESA-486AVR-128) take PC100 SDRAM, have onboard graphics, and a USB header. These can also take a mini-PCI card for SATA or gigabit ethernet. However this model uses a soldiered-in ZFx86 CPU with maximum 100 MHz.

Almost worth consideration if it wasn't over $300x10 and had 3 PCI, or PCI-E, ports. Actually, I'd even consider the ESA-MB103 model right now if the price was right (less than $50). I can get away the onboard IDE and an ISA NIC.

I'd be surprised if any of the late-model PCI-based 486's made it into mainstream industrial automation, let alone a PC Chips M919. It seems the main interest lies in ISA slots, where pre-PCI era 486 boards were loaded with ISA slots.

Plan your life wisely, you'll be dead before you know it.

Reply 41 of 280, by Tetrium

User metadata
Rank l33t++
Rank
l33t++

Bigger pic of the ALI board
http://www.tradenote.net/images/users/000/224 … ages/191496.jpg

And the complete list of that other manufacturers 486 boards. Now I want to see more 🤣!
http://www.esapcsolutions.com/industrial-moth … s-c-45_109.html

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 42 of 280, by feipoa

User metadata
Rank l33t++
Rank
l33t++

The ALi board seems to be substantially older -- the manual's date is 1999. I sent the company an e-mail inquiring about current pricing and availability. I asked for a quantity of 1 and for 5.

Plan your life wisely, you'll be dead before you know it.

Reply 43 of 280, by feipoa

User metadata
Rank l33t++
Rank
l33t++

There has been some talk of the POD100-WB in the other thread, although this probably belongs in this thread as it relates to the said scores.

To clarify, the only benchmark test that the POD100-WB didn't work with was the Quake Timedemo. Please read the footnote. The inserted value was linearly extrapolated from the data of POD83-WB, POD83-WT, and POD100-WT.

The working score of the POD100-WT in Quake is 23.6 fps.

Plan your life wisely, you'll be dead before you know it.

Reply 44 of 280, by Tetrium

User metadata
Rank l33t++
Rank
l33t++

It's a shame really that Intel never bothered to release a 100Mhz POD chip for Socket 3. Would've been awesome 😁

One thing that makes Socket 3 so interesting is that it's CPU's were being manufactured right to the end of the last millennium, even though for industrial purposes mostly. Those chips would be interesting to find out their overclocking potential 😁

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 45 of 280, by sliderider

User metadata
Rank l33t++
Rank
l33t++
Tetrium wrote:

It's a shame really that Intel never bothered to release a 100Mhz POD chip for Socket 3. Would've been awesome 😁

One thing that makes Socket 3 so interesting is that it's CPU's were being manufactured right to the end of the last millennium, even though for industrial purposes mostly. Those chips would be interesting to find out their overclocking potential 😁

What would have been better is if they just didn't lock the multiplier. There really was no reason to lock it. Officially, only the DX-50 motherboard or the 40mhz AMD setting would have overclocked it. Almost nobody knew about the undocumented bus speeds on some of the late PCI boards back then.

Reply 46 of 280, by Tetrium

User metadata
Rank l33t++
Rank
l33t++
sliderider wrote:
Tetrium wrote:

It's a shame really that Intel never bothered to release a 100Mhz POD chip for Socket 3. Would've been awesome 😁

One thing that makes Socket 3 so interesting is that it's CPU's were being manufactured right to the end of the last millennium, even though for industrial purposes mostly. Those chips would be interesting to find out their overclocking potential 😁

What would have been better is if they just didn't lock the multiplier. There really was no reason to lock it. Officially, only the DX-50 motherboard or the 40mhz AMD setting would have overclocked it. Almost nobody knew about the undocumented bus speeds on some of the late PCI boards back then.

True, but Intel decided to lock all of their overdrive processors (because Intel considered it's customers to be too stupid to know what a multiplier is? 😜). Shame they never released a non-overdrive Pentium for Socket 3 but I guess they only made these POD's because they had promised to do so

Edit:The whole point of the overdrive family was that you only needed to plug it in et voila, upgrade complete 😜

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 47 of 280, by DonutKing

User metadata
Rank Oldbie
Rank
Oldbie

The whole point of the overdrive family was that you only needed to plug it in et voila, upgrade complete

yeah, this was my understanding as well, and many 486 boards didn't have multiplier adjustment.

If you are squeamish, don't prod the beach rubble.

Reply 48 of 280, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Scans of the actual CPUs employed in this study have been added to the beginning of the thread. Scans were done using a retro UMAX Astra 600S SCSI scanner, purchased in 1997. The optics are starting to go as evident by the vertical/horizontal colour striations.

Plan your life wisely, you'll be dead before you know it.

Reply 49 of 280, by Tetrium

User metadata
Rank l33t++
Rank
l33t++
feipoa wrote:

Scans of the actual CPUs employed in this study have been added to the beginning of the thread. Scans were done using a retro UMAX Astra 600S SCSI scanner, purchased in 1997. The optics are starting to go as evident by the vertical/horizontal colour striations.

Very nice!!

And you got a Cyrix 5x86-80?? Never knew these existed!

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 50 of 280, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Actually, I have two! The only Cyrix 5x86 I'm missing is the 5x86-120/4X. I did not include duplicates of any CPU in this scan, even if the S-spec was different -- that would have made for a lot of Intel DX2's.

I have updated the PDF and text to reflect the comments and observations made by others. Thanks to everyone!

When I have the time (and motivation), I'll add the AMD DX4-100-WB-16KB to the list for improved clock-for-clock comparisons.

I cleaned up the CPU scan a bit to.

Plan your life wisely, you'll be dead before you know it.

Reply 51 of 280, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Attached is a PDF comparing the AMD DX4-100-WT-8KB vs. AMD DX4-100-WB-16KB. Significant improvements are highlighted. Some benchmarks show significant improvement, while most show negligable improvement (like Quake, Pcpbench, and 3Dbench).

The performance might match that of the Intel DX4-100-WB-16KB. I think I'll replace the AMD DX4-100-WT-8KB in the charts with the WB-16KB version and make a note of the overall differences. I can't see any reason to keep both when the results will be pretty similar, can you? The AMD DX4-120-WB-8KB will still outperform the DX4-100-100-WB-16KB.

Any thoughts?

EDIT: Bah, maybe I'll keep both on the chart. One reason I don't want to add another column is because it doesn't print as well due to scaling.

Attachments

  • Filename
    AMD_DX4-100_Comparison.pdf
    File size
    70.34 KiB
    Downloads
    385 downloads
    File comment
    Comparison of AMD DX4-100-WT-8KB vs. AMD DX4-100-WB-16KB
    File license
    Fair use/fair dealing exception

Plan your life wisely, you'll be dead before you know it.

Reply 52 of 280, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Here's a follow-up from the ESA motherboard.

ESA TF-486 (MB103) went non-production recently (less than a year ago), but there are a good 50K of them living out there. The reason they went non-production was because they couldn't get the ALI chipsets anymore. To continue prodution, they would have to put in a custom chipset order for 20K chipsets, which they decided not to do, or to buy used desoldered chipsets, which they decided not to do for reasons of contact oxidation, cleaning, prepping, and uncertainty in quality.

ESA no longer has an inventory of the TF-486. It was a great selling board. So the only other new option they offer is the ESA-486AVR-128, but you are stuck with the soldered-in custom processor. It does have a working USB port though.

Plan your life wisely, you'll be dead before you know it.

Reply 53 of 280, by feipoa

User metadata
Rank l33t++
Rank
l33t++

AMD DX4-100-WT and AND DX4-100-WB CPUs have been added to the official PDF and PNG images. The monologue has also been updated to reflect this change.

I suspect this will be the last required update.

EDIT: Bah, this is like writing a journal publication -- you're never really finished. I've added to the text, pdf, and png images data for an AMD X5-180, AMD X5-200, and a Cyrix 5x86-150. Values were linearly extrapolated from the data of AMD 66, 100, 133, and 160 and Cyrix 80, 100, 120, 133.

Plan your life wisely, you'll be dead before you know it.

Reply 54 of 280, by sebaz_ri

User metadata
Rank Member
Rank
Member

@feipoa
i think that the "Ultimate Socket 7 Benchmark" should be with the processors:
AMD K6
AMD K6-2
AMD K6-2+
AMD K6-3
Cyrix 6x86
Cyrix 6x86MX or MII
Intel Pentium MMX
Intel Pentium PRO
but... first we must overclock this processors up to the highest frequency

my records:
AMD K6-2 533 overclocked to 570 (95x6)
Cyrix M-II 233GP overclocked to 300MHz (100x3)
on a PCChips M598LMR board

Reply 55 of 280, by swaaye

User metadata
Rank l33t++
Rank
l33t++
sebaz_ri wrote:
@feipoa i think that the "Ultimate Socket 7 Benchmark" should be with the processors: AMD K6 AMD K6-2 AMD K6-2+ AMD K6-3 Cyrix 6 […]
Show full quote

@feipoa
i think that the "Ultimate Socket 7 Benchmark" should be with the processors:
AMD K6
AMD K6-2
AMD K6-2+
AMD K6-3
Cyrix 6x86
Cyrix 6x86MX or MII
Intel Pentium MMX
Intel Pentium PRO
but... first we must overclock this processors up to the highest frequency

Pentium Pro is a Socket 8 CPU.

Also,
Intel Pentium Classic (has half the L1 of PMMX)
Centaur Winchip / C6
Centaur Winchip 2
AMD 5k86 (SSA5)
AMD K5

Making a comprehensive Socket 7 comparison is going to be a lot of work...

Reply 56 of 280, by Tetrium

User metadata
Rank l33t++
Rank
l33t++
swaaye wrote:
Pentium Pro is a Socket 8 CPU. […]
Show full quote
sebaz_ri wrote:
@feipoa i think that the "Ultimate Socket 7 Benchmark" should be with the processors: AMD K6 AMD K6-2 AMD K6-2+ AMD K6-3 Cyrix 6 […]
Show full quote

@feipoa
i think that the "Ultimate Socket 7 Benchmark" should be with the processors:
AMD K6
AMD K6-2
AMD K6-2+
AMD K6-3
Cyrix 6x86
Cyrix 6x86MX or MII
Intel Pentium MMX
Intel Pentium PRO
but... first we must overclock this processors up to the highest frequency

Pentium Pro is a Socket 8 CPU.

Also,
Intel Pentium Classic (has half the L1 of PMMX)
Centaur Winchip / C6
Centaur Winchip 2
AMD 5k86 (SSA5)
AMD K5

Making a comprehensive Socket 7 comparison is going to be a lot of work...

And don't forget the Rise mP6 😉

And I reckon a good 2.2v sample MII chip could be overclocked to say 400Mhz.
And don't forget Tillamook (if we ever get the hecked chip running with L2 cache -_-)

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 57 of 280, by sliderider

User metadata
Rank l33t++
Rank
l33t++
Tetrium wrote:
And don't forget the Rise mP6 ;) […]
Show full quote
swaaye wrote:
Pentium Pro is a Socket 8 CPU. […]
Show full quote
sebaz_ri wrote:
@feipoa i think that the "Ultimate Socket 7 Benchmark" should be with the processors: AMD K6 AMD K6-2 AMD K6-2+ AMD K6-3 Cyrix 6 […]
Show full quote

@feipoa
i think that the "Ultimate Socket 7 Benchmark" should be with the processors:
AMD K6
AMD K6-2
AMD K6-2+
AMD K6-3
Cyrix 6x86
Cyrix 6x86MX or MII
Intel Pentium MMX
Intel Pentium PRO
but... first we must overclock this processors up to the highest frequency

Pentium Pro is a Socket 8 CPU.

Also,
Intel Pentium Classic (has half the L1 of PMMX)
Centaur Winchip / C6
Centaur Winchip 2
AMD 5k86 (SSA5)
AMD K5

Making a comprehensive Socket 7 comparison is going to be a lot of work...

And don't forget the Rise mP6 😉

And I reckon a good 2.2v sample MII chip could be overclocked to say 400Mhz.
And don't forget Tillamook (if we ever get the hecked chip running with L2 cache -_-)

Rise mp6 was only sold in Asia-Pacific markets so not many people outside that area ever got to use one. Even now it's hard to find one outside that area. The only seller that has them on ebay is the same one that sells the Tillamooks for $20 and his asking prices are outrageous. A real Pentium MMX or Pentium II of the same clock speeds is a lot cheaper and faster.

Reply 58 of 280, by feipoa

User metadata
Rank l33t++
Rank
l33t++

I currently have plans for a similar socket 7 comparison whereby all socket 7 CPUs will be tested on the same system. For completeness, I will include the PII-400, Cyrix 5x86, and POD results, in which case the results will no longer strictly be socket 7. A loosely translated title of this type of comparison may read, The Ultimate 586 Benchmark Comparison.

The CPUs in this study will include:

Cyrix
5x86 - 100 MHz (33 MHz x 3.0) for comparison against 6x86-100
5x86 - 120 MHz (40 MHz x 3.0) for comparison against 6x86-120
5x86 - 133 MHz (33 MHz x 4.0) for comparison against 6x86-133
6x86 - 75 MHz (50 MHz x 1.5)
6x86 - 90 MHz (60 MHz x 1.5) - U
6x86 - 100 MHz (66 MHz x 1.5) - U
6x86 - 120 MHz (60 MHz x 2.0)
6x86 - 133 MHz (66 MHz x 2.0)
6x86L - 150 MHz (60 MHz x 2.5)
6x86MX - 166 MHz (66 MHz x 2.5) - U
6x86MX - 200 MHz (66 MHz x 3.0)
MII - 233 MHz (66 MHz x 3.5) - U
MII - 250 MHz (83 MHz x 3.0)
MII - 266 MHz (66 MHz x 4.0) - U
MII - 300 MHz (75 MHz x 4.0)
MII - 333 MHz (83 MHz x 4.0) - O

Intel
P24T - 83 MHz (33 MHz x 2.5)
P24T - 100 MHz (40 MHz x 2.5) - O
P54C - 75 MHz (50 MHz x 1.5)
P54C - 90 MHz (60 MHz x 1.5)
P54C - 100 MHz (66 MHz x 1.5)
P54CS - 120 MHz (60 MHz x 2.0)
P54CS - 133 MHz (66 MHz x 2.0)
P54CS - 150 MHz (60 MHz x 2.5) - U
P54CS - 166 MHz (66 MHz x 2.5)
P54CS - 200 MHz (66 MHz x 3.0)
P55C-MMX - 100 MHz (?) - U
P55C-MMX - 166 MHz (66 MHz x 2.5) - U
P54C-MMX - 200 MHz (66 MHz x 2.5) - U
P54C-MMX - 233 MHz (66 MHz x 3.5)
P54C-MMX - 266 MHz (66 MHz x 4.0)
P54C-MMX - 300 MHz (75 MHz x 4.0) - O
P54C-MMX - 333 MHz (83 MHz x 4.0) - O
PII - 400 MHz (100 MHz x 4.0)

AMD
K5 - 100 MHz (66 MHz x 1.5)
K6-2 - 100 MHz (?) - U
K6-2 - 266 MHz (66 MHz x 4.0)
K6-2 - 300 MHz (75 MHz x 4.0) - O
K6-2 - 500 MHz (83 MHz x 6.0) - U
K6-3+ - 400 MHz (66 MHz x 6.0) - U
K6-3+ - 450 MHz (75 MHz x 6.0)
K6-3+ - 500 MHz (83 MHz x 6.0) - O

IDT
Winchip - 200 MHz (66 MHz x 3.0)
Winchip2 - 233 MHz (66 MHz x 3.5)
Winchip2 - 266 MHz (66 MHz x 4.0) - O

In cases where I do not have the actual CPU at the specified frequency, a closely matched CPU will be benchmarked in its place (slightly over/under clocked, provide that the CPU architecture is the same). U stands for an underclocked CPU while O stands for an overclocked CPU. The clock frequencies noted are actual clock frequencies, not the PR rating using by IDT, Cyrix6x86/MII, and AMD K5's. For example a Cyrix MII-400GP will be used for the 300 MHz test.

It will be interesting to do a clock-for-clock comparison at 100 MHz with POD100, Cyrix 5x86-100, K5-100, Intel100, Cyrix 6x86-100, and perhaps too the the K6-100 and MMX-100 if they will downclock well.

Is there any CPU of particular interest I have left out? I can only go up to 500 MHz for AMD and 333 MHz for Cyrix. If the Intel CPU's are locked, then I am stuck with a 4x multiplier, for which 333 MHz will be the max I can test on this non-super socket7. The BIOS has been modified to allow for proper K6 II/III+ operation as well as Cyrix 4x, with a maximum FSB of 83.3 MHz.

Last edited by feipoa on 2011-12-27, 11:06. Edited 1 time in total.

Plan your life wisely, you'll be dead before you know it.