VOGONS


Bitchin' dual p-pro setup

Topic actions

Reply 41 of 84, by luckybob

User metadata
Rank l33t
Rank
l33t

If you have a better idea, im all ears. Now is the time to find this stuff out properly. Once i load everything down with 20+ games, things might be different.

It is a mistake to think you can solve any major problems just with potatoes.

Reply 43 of 84, by dirkmirk

User metadata
Rank Oldbie
Rank
Oldbie

Any chance of a 3dmark 2001 benchmark with a dx9 card? I suppose any of the Geforce 5 series or ATI 9000s will max the score out, I'm getting around 1,800 on my machine.

Reply 44 of 84, by luckybob

User metadata
Rank l33t
Rank
l33t

A dx9 card on PCI is going to be a challenge. The fastest pci card I own is the ati 9250. And I have yet to find it. I will happily oblige if I can get the card found.

In other news, I did some more testing. I was faffing about with performance test 6.1, which is a newer version of passmark 4.0 if i'm correct. It had three "advanced" presets for hard drive testing. workstation, web server, & file server. Each basically set a slider defining % read to write, random % and such. No test gave any sort of conclusive results that I could measure. I then decided to run all three tests at once and a 4th thread on a network drive. Hoping I could just saturate the damn thing and get some results. All tests were done with the onboard scsi and the 15krpm scsi drive I have. The network was limited to a 10/100 as I have yet to get a gigabit card that works with pci 2.1.

images: http://imgur.com/a/6w2eb

some reason the 3rd thread got almost zero in the scores. If I ran just one test, I get the highest mark almost exactly. I'm probably not doing something right. any advice?

It is a mistake to think you can solve any major problems just with potatoes.

Reply 48 of 84, by luckybob

User metadata
Rank l33t
Rank
l33t

finally found a working gigabit Ethernet card. It is a stock Intel one. had to force it slightly into the pci slot because the card itself is pci-x. Made no difference in the previous test. however I did find a client?server benchmark in this suite. so I installed it on my 2x quad core xeon monster and had a run with that.

I tested several 6 different ways. send, receive, and 2 threads. cached and uncachced for each. Well, the only concrete results i got were the i960 caps out at 70mb/s. when in bridge mode, I can cap out the pci bus at ~93mb/s. It is slightly interesting how the 2 threads took 2 cpus to 100% each. and on my xeon the test didn't even register on the cpu usage meter. 😜

the testing continues!

It is a mistake to think you can solve any major problems just with potatoes.

Reply 49 of 84, by 386_junkie

User metadata
Rank Oldbie
Rank
Oldbie
luckybob wrote:

the testing continues!

That's super!

Compaq Systempro; EISA Dual 386 ¦ Compaq Junkiepro; EISA Dual 386 ¦ ALR Powerpro; EISA Dual 386

EISA Graphic Cards ¦ EISA Graphic Card Benchmarks

Reply 50 of 84, by luckybob

User metadata
Rank l33t
Rank
l33t

I have found a 2nd board!

Well kinda. It is actually a Supermicro P6DLH. just won it yesterday. http://www.ebay.com/itm/301969642018

now this wont be a dual socket 8, but dual slot 1. The auction comes iwht a pair of 333's! I did not have a pair of these chips, so I can scratch 2 things off the list. The only difference between the p-pro overdrive and real p2-333's is the L2 cache. The overdrives have fullspeed cache, instead of the 1/2 speed of the "consumer" P2.

So I can now do a bit of testing betweent the chips. the slower cache is a negative, BUT it might be offset by the faster main memory (sdram). YAY! Isn't this fun?

It is a mistake to think you can solve any major problems just with potatoes.

Reply 51 of 84, by chinny22

User metadata
Rank l33t++
Rank
l33t++

Dual CPU's = fun
Slot CPU's = fun

luckybob wrote:

Isn't this fun?

Fun doesn't cut it anymore! I'm not sure this amount of enjoyment is even legal!!!

Reply 52 of 84, by vetz

User metadata
Rank l33t
Rank
l33t

Very nice! I know you've been searching along time for this board!

Have the UP5 myself with dual socket 7 or dual Pentium Pro and 512mb RAM. Finding an AT case for UP5 was hard, but in the end I managed to mount it in an ATX case without modding.

3D Accelerated Games List (Proprietary APIs - No 3DFX/Direct3D)
3D Acceleration Comparison Episodes

Reply 53 of 84, by luckybob

User metadata
Rank l33t
Rank
l33t

Got another motherboard recently. A Supermicro P6DLH. VERY similar to the asus. But overall a board that has several pros and cons compared to the P65UP8. Basically it comes down to how you want to use the board.
edit pic:
tZU7Uqlm.jpg

Different pci placement, no onboard scsi, daughtercard, chipset, overclocking!

I haven't played much with the new board much, but it seems to work fine at first glance. It supports overclocking! the board has silk-screened options for 50 to 75mhz fsb. I googled the clockgen chip and it has MANY more options. Most notably it will go up to 83mhz fsb. This works fine on the P6DLH, but the overclocking is disabled on the P65UP8. Same chip. I was REALLY hoping I could overclock the p-pro OD chips. the P65UP8 has the same clockgen, but I suppose the daughtercard overrides the jumper settings somehow. Or the OD chips set it to 66mhz, or both. Anyway, I did a few basic tests and they came out exactly as you would expect a p2-333 would be. it does work just fine @ 83mhz! This new board has the bios option for USWC, but I only get ~17fps in quake. So yea, I do need to investigate this.

It is a mistake to think you can solve any major problems just with potatoes.

Reply 55 of 84, by luckybob

User metadata
Rank l33t
Rank
l33t
mrau wrote:

how fast are these OD's compared to normal p2?
is the graphics card on the i960 or the other ports?

significantly. The core is identical, but the overdrives have full speed L2 cache, where the normal p2 uses 1/2 speed.

I have the gfx card on the normal pci bus.

It is a mistake to think you can solve any major problems just with potatoes.

Reply 56 of 84, by mrau

User metadata
Rank Oldbie
Rank
Oldbie
luckybob wrote:

significantly.

yes yes 😀 but any tests that you have done that would allow for direct comparison?

luckybob wrote:

I have the gfx card on the normal pci bus.

did You try those special cpu tools that are usually mentioned here when someones doing and amd build?

Reply 57 of 84, by luckybob

User metadata
Rank l33t
Rank
l33t

no, not yet. I do have it on my short list of things to do.

what cpu tools are you referring to? Also, this is Intel, so I hadn't thought to even look at them if I'm honest.

It is a mistake to think you can solve any major problems just with potatoes.

Reply 58 of 84, by mrau

User metadata
Rank Oldbie
Rank
Oldbie

MTRRLFBE http://rayer.g6.cz/programm/programe.htm this or similar ones; im not sure if this one works with p6, it shouldnt set your box ablaze either way

more's mentioned here https://en.wikipedia.org/wiki/Write_combining

Reply 59 of 84, by dirkmirk

User metadata
Rank Oldbie
Rank
Oldbie

I've had success increasing the dos video performance on my PR440FX @ 350mhz, I switched the video card to an Nvidia FX5500 128bit 256MB.

The quake demo3 performance was only 16.1fps without fastvid running 640x480 fullscreen minus 2, running fastvid the score increases to 28.6 it runs very nice despite the resolution it looks sharp unlike most accelerated games appear that muddy(at 640x480).

Speedsys

Memory EDO DIMM - 256mb
Memory Bandwidth - 263.88mb/s
Video - 79,327kb/s
CPU score - 402.21
Datacache L1 (16Kb) - 1331.68mb/s
Datacache L2 (512Kb) - 495.11mb/s
Memory throughput - 138.41mb/s

Quake 640x480 - minus 1 fullscreen

demo1 - 29.2fps (80.3fps @ 320x200)
demo3 - 27.7fps

Pcpbench 640x400 - 49.1fps

3Dbench 2 - 469.2fps

Does it look like memory is holding the system back? I'm not sure what this machine should be benching.