4000 Cpu

[eschw95458]

Ok, I managed to upgrade the cpu in my 4100 from 233 to 400. Now it SEEMS to be faster, but the real question is how do we tell if it really is faster, or just running at the same Mhz as the old one.

[blue68f100]

FreeBSD logs the hardware on startup. But I don't know if the snap does. Never have looked for it. The snap shell keeps you out the system files. With FreeBSD you hit the "scroll lock key" once. Then page up or arrow up to go through the log. Then "Scroll lock key" again to drop out to cmd prompt.

[re3dyb0y]

Heh

Seeming could be just you thinking/hoping it is....

In theory boot up time should be a bit better

[jontz]

I would imagine that the processor speed is set on the mobo with jumpers like most older boards. In the snaps case, they are probably solder bridges rather than jumpers you can move. My guess is that you are running at the same clock speed as before...

[eschw95458]

No, no real jumpers. From what I have read from intel's website on this particular chipset, the manufacturer could set the cpu speed or leave the chipset in Auto, which gets its info from the cpu. It is really hard to say if Quantuim would have taken the extra effort to lock the chipset into one speed, one would believe that would be an unnecessary extra step.
Now having said all that, if nothing else the amd cpu has more cash on the chip compaired to the intel.
If anyone wants that document let me know.

[Hallis]

Originally Posted by eschw95458

No, no real jumpers. From what I have read from intel's website on this particular chipset, the manufacturer could set the cpu speed or leave the chipset in Auto, which gets its info from the cpu. It is really hard to say if Quantuim would have taken the extra effort to lock the chipset into one speed, one would believe that would be an unnecessary extra step.
Now having said all that, if nothing else the amd cpu has more cash on the chip compaired to the intel.
If anyone wants that document let me know.

But im wondering what the real-world benefits are in terms of RAID rebuild speed, etc.

Shane

[blue68f100]

Originally Posted by eschw95458

No, no real jumpers. From what I have read from intel's website on this particular chipset, the manufacturer could set the cpu speed or leave the chipset in Auto, which gets its info from the cpu. It is really hard to say if Quantuim would have taken the extra effort to lock the chipset into one speed, one would believe that would be an unnecessary extra step.
Now having said all that, if nothing else the amd cpu has more cash on the chip compaired to the intel.
If anyone wants that document let me know.

Look at the pin out on the CPU. There should be one pin that is required to be low or High. See if that pin goes anywhere.

[jontz]

Originally Posted by eschw95458

No, no real jumpers. From what I have read from intel's website on this particular chipset, the manufacturer could set the cpu speed or leave the chipset in Auto, which gets its info from the cpu. It is really hard to say if Quantuim would have taken the extra effort to lock the chipset into one speed, one would believe that would be an unnecessary extra step.
Now having said all that, if nothing else the amd cpu has more cash on the chip compaired to the intel.
If anyone wants that document let me know.

Cool info! I'll take a copy of that document...my email is in my profile.

[eschw95458]

Ok, did some testing last night, I have a 4000 with 128m 400mhz cpu. Boot time is app. 30-35 sec from dead stop to browser refresh. 4100 with 128m 233mhz takes 65-70 sec boot.
Can someone with a 4000 with 128 mem or better check their boot time. Type http://snapserver/config into browser, press start on snap, when fans kick on, start count. About ten seconds after fans start, hit enter or reload the browser, when the page finally refreshes, that is the count I am using. (time the fans start to page refresh)

Few other thinigs I can tell you about the 4000/4100. Both use the 82439tx Northbridge chip, Both use pentium pro 233 mhz cpu.

[Hallis]

I'll do a bootup test on my 4000 when i get home tonight. It should be 100% bone stock.

Shane

[jontz]

Originally Posted by eschw95458

Ok, did some testing last night, I have a 4000 with 128m 400mhz cpu. Boot time is app. 30-35 sec from dead stop to browser refresh. 4100 with 128m 233mhz takes 65-70 sec boot.
Can someone with a 4000 with 128 mem or better check their boot time. Type http://snapserver/config into browser, press start on snap, when fans kick on, start count. About ten seconds after fans start, hit enter or reload the browser, when the page finally refreshes, that is the count I am using. (time the fans start to page refresh)

Few other thinigs I can tell you about the 4000/4100. Both use the 82439tx Northbridge chip, Both use pentium pro 233 mhz cpu.

That's a heck of an improvement! I need to get another snap to play with. I don't like messing with my "production" 4100...

[Hallis]

Originally Posted by jontz

That's a heck of an improvement! I need to get another snap to play with. I don't like messing with my "production" 4100...

Yah, As far as production units go, If it ain't broke, don't fix it.

Shane

[blue68f100]

Let's see some bench marks on file transfers, read and write. We know the raid 5 can max the 100base network. But we do not know what kind of performace it will do in RAID 1. Compaired to a 2200 which seams to limit at ~4.3 Meg/sec Write speed. And it is more that twice as fast as my 2000 before it had a painful death.

[Hallis]

Originally Posted by blue68f100

Let's see some bench marks on file transfers, read and write. We know the raid 5 can max the 100base network. But we do not know what kind of performace it will do in RAID 1. Compaired to a 2200 which seams to limit at ~4.3 Meg/sec Write speed. And it is more that twice as fast as my 2000 before it had a painful death.

I'm wondering if the hard drives themselves will also effect the speed of transfers by speed, ammount of cache, ect. So different people might have different results.

It'd also take longer to transfer 100mb of small files then it would to transfer 1x 100mb file. I think there are still too many variables in this. I would advise that one person do all the testing so that the only thing changing is the CPU.

BTW David i might want to get that 2000 from you and use it for parts or try to get it working.

Shane

[eschw95458]

Ok, made Raid one config, two drives (Seagate 7200.10) 130G(for now) made one file @ 100mb
and one movie file @1g. Up speed to snap shows 6.9-7.0. Down from snap shows 10.6

Specs are
Snap 4000 (400cpu)
ver 3.4.803
2 x Seagate 7200.10
128 mem
HW 2.0.1
Bios 2.0.252

[jontz]

Originally Posted by Hallis

BTW David i might want to get that 2000 from you and use it for parts or try to get it working.

Shane

That's his project for a rainy day, methinks...

[eschw95458]

Actually, I did work on it, I got the same results with the slower processer. Still too many variables.
Is this particular test really using much of the cpu cycle? Have they (Quantum) used a kernel optimized for the intel cpu? Or is the new cpu just running at the same speed as the old one? I'm open to suggestions.

[jontz]

I don't know that reading from the snap uses much of the CPU, writting to the snap would be the more processort intensive task. I still tend to think that the new CPU is running at the same speed as the old one, but I have no way to prove that.

The other thing that comes to mind is that we will have no real way to test performance on a 100Mb network. I think it would have to be a 1Gb to really know what is going on, since one of our limiting factors is the network at this point.

[re3dyb0y]

CPU Intensive would be something java based

Ram as well, of course,,,