VOGONS


Powerleap adapter

Topic actions

Reply 60 of 77, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

I believe back in those days chip companies had to make a choice between chips that could get a lot of work done per clock, but not clock as high or a chip that was slower per clock, but could be clocked very high...

The legendary 6x68 is one example. It was faster per clock than Intels Pentium Something very very hard to pull off. However they couldn't clock it much higher and that was pretty much the end for Cyrix.

From what I read, the PIII was similar as Intel struggled to get past 1 GHz and keep up with AMD. So for them P4 was the answer as they could easily go past 1 GHz and where back in the race. There was a PIII 1.13 GHz I believe which was so unstable that Intel had to recall it...

So back then it was pretty much all about the GHz race...

Reply 61 of 77, by swaaye

User metadata
Rank l33t++
Rank
l33t++

Yup the clock speed that a chip can manage is determined by many things. Some chips did a lot per clock but couldn't get much clock speed so it didn't matter.

AMD K5 and 6x86 are good examples. K6 is a good one too because even on 180nm it can't do much more than 600 MHz whereas an Athlon or P3 can do over 1 GHz.

AMD K5 is actually one of the fastest x86 CPUs on a per clock level. You can see this in my "133 MHz Challenge" link where all the CPUs are at 133 MHz. It is very competitive. But obviously Intel didn't stop at 133 MHz so K5 became irrelevant quick.

Reply 62 of 77, by retro games 100

User metadata
Rank l33t
Rank
l33t

I've been thinking about benchmarking in general, and wondering what is a good way to test the processing power of an old machine. I wonder if SuperPi would be a worthy candidate for this task? Vogons user Tetrium started a thread here about this utility, and it contains a download link too.

It takes quite a while to run the "full test", which is called 1M, on a 486. Perhaps a more practical approach would be to select the 16K test, if running on a 386 for example.

Mau1wurf1977, you may wish to consider adding SuperPi to your benchmarking statistics. And also, what would be really useful is if someone out there could also run SuperPi on an actual 386 machine - preferably of the DX variety, eg a 33 MHz model.

Reply 63 of 77, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

I can do that! Have only tested a few cpus.

Where do I donwload it? What version should I use? And is it DOS / is there a boot CD version (linux or something)

EDIT: Having second thoughts because I don't remeber super PI being a popular benchmark in the 386/486/Pentium days...

Reply 64 of 77, by retro games 100

User metadata
Rank l33t
Rank
l33t

OK. But if you change your mind, the download you need can be found in Tetrium's Vogons topic, the link to which can be found in my last post, above. The download version you want is the Win9x version.

Reply 65 of 77, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

Cool!

My tables will use results from:

NORTON SI 8.0
SPEEDSYS 4.78
3DBENCH v1.0 / 1.0c (2)
PCPBENCH (VGAMODE) (3)
PCPBENCH

(2) I found v1.0 more precise for slow machines (<100 fps). For fast machines (>100 fps) use Version 1.0c
(3) Launch benchmark via PCPBENCH /VGAMODE

PS: My Super Socket 7 boards still haven't arrived 🙁

Reply 66 of 77, by retro games 100

User metadata
Rank l33t
Rank
l33t
Mau1wurf1977 wrote:

PS: My Super Socket 7 boards still haven't arrived 🙁

Oh dear. 🙁 What is the shipping country of origin - is it far away? I remember a few international parcels taking some time to arrive. Two reasons: the seller sent the items by surface and not by air, and also some parcels were inspected at customs, and an import duty was subsequently imposed on them. That can waste a bit of time!

Reply 67 of 77, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

The boards are from Australia, I guess I just need to relax. The CPUs only arrived yesterday and I got them at a similar time...

I am bored of benching Slot 1 stuff 😀

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

Reply 68 of 77, by retro games 100

User metadata
Rank l33t
Rank
l33t

Introduction
My Tualatin SL6BY processor has arrived. Its specification is P3-S Tualatin 1.4GHz 512 cache 133FSB. Using the same test mobo as before, I'm going to race it against two other Pentium 3 processors: 1200 MHz and 1400 MHz Celeron models. These Celerons are both 100 FSB models. Luckily the 1200 MHz version overclocks well, using an FSB of 133. However, the 1.4 GHz Celeron can only tolerate an FSB overclock of 120 FSB.

Test hardware
It hasn't changed. I'm still going to be using the Asus P2B mobo, with a Powerleap adapter. The video card hasn't changed either, it's still the GF3 Ti200.

Test apps
Previously, when I tested the Celeron 1200 and 1400 processors, I used Sandra 2002 Pro and 3DMark 2000. I'll use the same apps again for the new Tualatin processor tests.

Test results
The best score is in the final right column. The CPU MHz column heading of "1574" means that I am testing the Tualatin 1.4GHz processor, and overclocking the motherboard to 150 FSB.
tualatinResults.jpg

150 FSB screenshot
The following screenshot shows the system running with the 1.4GHz 133FSB Tualatin CPU, and the motherboard overclocked to 150 FSB. On the left hand side is Sandra 2002 Pro showing information about the CPU, and on the other side is 3DMark 2000 showing its completed score.
tualatin150.jpg

Last edited by retro games 100 on 2010-10-01, 06:42. Edited 1 time in total.

Reply 69 of 77, by swaaye

User metadata
Rank l33t++
Rank
l33t++

I managed to run a Katmai P3 on my Abit BF6 440BX mobo on a 150 MHz FSB way back in the day. 😁 That put it up around 675 MHz. I was really pumping the volts through that thing. It wasn't exactly stable but I was playing Quake 3 and it was working ok for a bit.

By forcing the Matrox G400 to AGP 1x it was somewhat happy with that 100 MHz AGP.

Reply 70 of 77, by Tetrium

User metadata
Rank l33t++
Rank
l33t++
Mau1wurf1977 wrote:

I can do that! Have only tested a few cpus.

Where do I donwload it? What version should I use? And is it DOS / is there a boot CD version (linux or something)

EDIT: Having second thoughts because I don't remeber super PI being a popular benchmark in the 386/486/Pentium days...

Thanks for referring to my superpi thread! Haven't posted results in quite a while since I've build no new rigs for a while now.

Maulwurf, I picked superpi as it allows me to directly compare the widest amount of systems with eachother which is what I wanted. To be able to put the 486 and Athlon 64 and beyond right next to eachother!

I'd prefer such a program bootable from a simple DOS bootdisk but haven't found the perfect bench program so far.

Reply 71 of 77, by archsan

User metadata
Rank Oldbie
Rank
Oldbie
retro games 100 wrote:
Test hardware It hasn't changed. I'm still going to be using the Asus P2B mobo, with a Powerleap adapter. […]
Show full quote

Test hardware
It hasn't changed. I'm still going to be using the Asus P2B mobo, with a Powerleap adapter.

150 FSB screenshot
The following screenshot shows the system running with the 1.4GHz 133FSB Tualatin CPU, and the motherboard overclocked to 150 FSB. On the left hand side is Sandra 2002 Pro showing information about the CPU, and on the other side is 3DMark 2000 showing its completed score.
tualatin150.jpg

Wow, RG100, the ASUS P2B is doing well up to 150MHz FSB? 😮
And you're still using the Voodoo3 AGP card?

I think i should get a 1.4 Tualatin to try it myself. Is the SL6BY the best one to get?

Thanx

Reply 72 of 77, by retro games 100

User metadata
Rank l33t
Rank
l33t

Thanks very much people, for everyone's really interesting comments so far.

Re: pumping lots of volts through a Katmai, back in the day. Y'know ebay is awash with processors, and let's all take a moment to thank ebay for being able to get this retro stuff, at such low prices! Back in the day, a Katmai can't have been cheap. Swaaye, were you a bit nervous about OC'ing that chip to 150MHz?! 😮

Re: SuperPi. Hehe, I tried the 16K test using a P3 system with disabled cache. It seemed to work just fine. The only slight issue was it takes a long time to boot in to Windows 98. At least 5 minutes in fact!

Re: testing hardware. Archsan, well spotted. I didn't make it clear in my previous post that I was still using the GF3 Ti200 AGP video card. I will edit my post above to reflect this information. (Please note that I'm not using a V3 card ATM.) Also, the 1.4 GHz Tualatin seems to be the best processor to run games. As soon as you benchmark it using 3DMark, it beats all other processors.

I'm both surprised and impressed that the Asus P2B can handle 150 FSB. Perhaps it's a combination of the board revision (1.12), and the fact that I'm using a Powerleap adapter. It does seem to be a good piece of hardware.

Reply 73 of 77, by buckrogers

User metadata
Rank Newbie
Rank
Newbie
archsan wrote:

Wow, RG100, the ASUS P2B is doing well up to 150MHz FSB? 😮
And you're still using the Voodoo3 AGP card?

I think i should get a 1.4 Tualatin to try it myself. Is the SL6BY the best one to get?

The SL6BY is the Manny Pacquiao of P3 cpu's...

http://en.wikipedia.org/wiki/Manny_Pacquiao

Reply 74 of 77, by retro games 100

User metadata
Rank l33t
Rank
l33t

I decided to see how stable the system was at 150 FSB. Inside the BIOS set up options area, I set the timings (including memory) to aggressively fast settings. I then reran 3DMark 2000. It worked, and I improved the previous score of 9312 to 9456. I then overclocked the GF3 Ti200 video card from its default setting of 175/400, to 210/491 (which is just under the "safe max" level), and reran 3DMark 2000. It worked, and I improved the score of 9456, to a very respectable 10779. The following screenshot shows this "hot rodded" information.

150.jpg

Reply 75 of 77, by buckrogers

User metadata
Rank Newbie
Rank
Newbie

RG100, keep up the good work. If no one else, then I'll at least be drawing on this knowledge when I get round to putting together my own P2b based system.

Reply 76 of 77, by retro games 100

User metadata
Rank l33t
Rank
l33t

I wanted to run the DOS VGA benchmarking tool 3DBench version 1.0 again. This time, the "hardware profile" is: Asus P2B, 150 FSB, 1.4 GHz Tualatin, Riva 128 4MB PCI card, mobo BIOS settings "maxed out", L1 cache disabled.

I have now improved my score from 14.7 to 16.3. I'm happy with that! Also, the PCI video card appears to be comfortable running on the OC'd PCI bus. As the mobo's FSB is being OC'd to 150 FSB, the PCI bus is now running at 37.5 FSB.

Reply 77 of 77, by swaaye

User metadata
Rank l33t++
Rank
l33t++
retro games 100 wrote:

Re: pumping lots of volts through a Katmai, back in the day. Y'know ebay is awash with processors, and let's all take a moment to thank ebay for being able to get this retro stuff, at such low prices! Back in the day, a Katmai can't have been cheap. Swaaye, were you a bit nervous about OC'ing that chip to 150MHz?! 😮

I can't remember how much it cost... It was a P3 450 SL35D which was known to very frequently have the same cache chips as the 600 MHz model. I didn't get it until mid '99. It was an upgrade from my P2 300 SL2W8 which had the cache chips of a 450.

Katmai runs 2.0 volts like Deschutes so I couldn't push it all that far on the overvolting. I think 2.4v is the limit of my BF6 (I'm still using that board today!) I had a pair of massive Slot 1 CPU coolers for my crazy tweaking as well.

I also can't remember what happened to either of those CPUs. 😁 I bought a Coppermine towards the end of '99 and the oldies went somewhere..... Actually, I bought another SL35D on ebay a couple of years ago and it again does 600/133 no prob. That is a PIII 600B in Intel's P3 naming scheme of the day.

I think that if Athlon had been delayed that Katmai would have been around longer. It felt like Athlon forced their hand.