VOGONS


First post, by noshutdown

User metadata
Rank Oldbie
Rank
Oldbie

all using 50fsb

pentium has a minimum multiplier of 1.5x, so its pr75, which is used for reference.
k6-2/3 series have a minimum multiplier of 2.5x so they are out of consideration. k6-3 multiplier can be set to 2x using software and get 100mhz, but is still in no way slower than reference.
k6 also has minimum multiplier of 2x, and i would rate it on par with pr75 reference.
k5 has low clock and few multiplier settings: 1.5x, 1.75x and 2x, i dunno if there are other hidden settings. if 1.5x is minimum then k5-75 is just slightly slower than reference and i would rate it pr66.

now for the slowest clock-to-clockers: idt c6 and cyrix 6x86, they both appears to support only integer multipliers, and i dont have plenty of info on how they behave.
idt c6: the slowest clock to clock, if 2x is minimum then its 100mhz and i would rate it pr50, slower than the k5-75.
cyrix 6x86: this may be the deciding factor. its a bit faster than idt c6 clock-to-clock, but rumored to run at 1x when jumpers are 2.5x (i have no info on this). if this true then its the only cpu than run 1x on socket7, and i would rate it pr30.

conclusion:
if 6x86 can run at 1x then its the "winner"
else, if k5 has hidden multipliers lower than 1.5x then it wins.
or else, idt c6 wins.

Last edited by noshutdown on 2012-06-18, 02:42. Edited 1 time in total.

Reply 2 of 26, by feipoa

User metadata
Rank l33t++
Rank
l33t++

The Cyrix 6x86 supports 1x multipliers, or at least the original Cyrix 6x86-80 I tested did. I had to set the motherboard to 4x to get the CPU to run at 1x. I ran it at 1.0 x 75 and 1.0 x 83 as part of the Ultimate 6x86 Benchmark Comparison (in progress / on hold). So if you ran it at 1.0 x 50 MHz, I think this would be the slowest socket 7 CPU configuration available without modifying the motherboard's clock generator for sub-50 MHz speeds.

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

Reply 3 of 26, by leileilol

User metadata
Rank l33t++
Rank
l33t++

A 6x86-80 is still damn speedy at Zdoom (i've tested a Cyrix 6x86MX at that speed before years ago), so slowest FPU it isn't IMO

of course I could (and should) be proven wrong, I believe K5 could take the 'slowest s7' cake.

apsosig.png
long live PCem

Reply 4 of 26, by feipoa

User metadata
Rank l33t++
Rank
l33t++

80 MHz is quite a bit faster than 50 MHz. My understanding is that we aren't looking for the slowest clock-for-clock socket 7 FPU, but the slowest FPU possible on a socket 7. The main take-away message as that the 6x86 can run at 50 MHz. Can the K5, or others, run at 1.0X at 50 MHz? The original 6x86 is non-MMX.

Is zDoom highly FPU intensive? The original DOOM was mainly, or entirely, ALU demanding, though I'm not familiar with testing zDoom.

If there is enough interest in this, I could do a slow-moe show-down between the Cyrix 6x86-50 and any other socket 7 you think can run at 50 MHz.

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

Reply 5 of 26, by nforce4max

User metadata
Rank l33t
Rank
l33t
feipoa wrote:

The Cyrix 6x86 supports 1x multipliers, or at least the original Cyrix 6x86-80 I tested did. I had to set the motherboard to 4x to get the CPU to run at 1x. I ran it at 1.0 x 75 and 1.0 x 83 as part of the Ultimate 6x86 Benchmark Comparison (in progress / on hold). So if you ran it at 1.0 x 50 MHz, I think this would be the slowest socket 7 CPU configuration available without modifying the motherboard's clock generator for sub-50 MHz speeds.

I've gone lower but most boards don't support lower clocks and the few that do don't advertise them. I did manage to get a p1 to run around 20mhz or so and took close to a min just to post. I discovered the setting when trying to figure if it supported higher fsb clocks on a older oem Gateway board (socket 5) that was using the Intel FX series chipset which was before HX and VX era. Sadly it was one of the boards that I lost and worse a K6 with the big logo was installed just before the move 😢

The oddest board that I have worked with besides a Media GX build was a Compaq that used the rare Opti Viper chipset.

On a far away planet reading your posts in the year 10,191.

Reply 6 of 26, by Filosofia

User metadata
Rank Member
Rank
Member

I'm currently running a 240MHz rated C6 at 133MHz on my dos gaming machine (with FSB slightly overclocked to 66MHz , motherboard does not support 60FSB or is not documented) and have a Cyrix PR200 that posted PR75 too (I don't have a clue FSB or multiplier) on the same board.
Maybe I'll measure the FSB scalling for both.

Reply 12 of 26, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

Per HDMI bekommt man glaube ich automatisch diese frequenzen. Schwierigier ist es per DVI. Da muss man eigene Aufloesungen erstellen.

My website with reviews, demos, drivers, tutorials and more...
My YouTube channel

Reply 13 of 26, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

Aren't there boards that support 40MHz bus?

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 14 of 26, by feipoa

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

How would you test FPU performance alone? SuperPi?

What about Landmark? It has seperate scores for ALU and FPU.

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

Reply 15 of 26, by elianda

User metadata
Rank l33t
Rank
l33t

What about the classic Whetstone Benchmark?
http://www.roylongbottom.org.uk/whetstone%20results.htm

Retronn.de - Vintage Hardware Gallery, Drivers, Guides, Videos. Now with file search
Youtube Channel
FTP Server - Driver Archive and more
DVI2PCIe alignment and 2D image quality measurement tool

Reply 16 of 26, by noshutdown

User metadata
Rank Oldbie
Rank
Oldbie
Anonymous Coward wrote:

Aren't there boards that support 40MHz bus?

maybe they exist but i havn't seen any. after all, there seemed to be no pentium or k5 officially rated for 40fsb, only cyrix has one model rated 40*2.

Reply 17 of 26, by Filosofia

User metadata
Rank Member
Rank
Member
elianda wrote:

What about the classic Whetstone Benchmark?
http://www.roylongbottom.org.uk/whetstone%20results.htm

Winchip C6 200MHz 40.6 16.9 12.3 7.81 1.02 0.71 18.4 28.4 17.0
Pentium 200 200MHz 132.0 54.2 34.7 20.3 5.12 3.03 32.5 39.8 46.2

😈

Reply 18 of 26, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

I know of one board that does 40MHz: Shuttle HOT 553

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 19 of 26, by noshutdown

User metadata
Rank Oldbie
Rank
Oldbie
Filosofia wrote:
Winchip C6 200MHz 40.6 16.9 12.3 7.81 1.02 0.71 18.4 28.4 17.0 Pentium 200 200MHz 132.0 54.2 34 […]
Show full quote
elianda wrote:

What about the classic Whetstone Benchmark?
http://www.roylongbottom.org.uk/whetstone%20results.htm

Winchip C6 200MHz 40.6 16.9 12.3 7.81 1.02 0.71 18.4 28.4 17.0
Pentium 200 200MHz 132.0 54.2 34.7 20.3 5.12 3.03 32.5 39.8 46.2

😈

this benchmark is obvious theoretical. that is, only the sole fpu and no cache/ram were involved, which would not correctly represent the actual performance.
for example, celeron733 scores higher than pentium3-700, which is undoubtedly a joke. all pentium4 series score incredibly low in the table aswell, their real life performance may be bad but not to that extend.
for me, superpi is a more realistic fpu&cache&ram benchmark, and is more related to actual performance.