New core speed relative to old core speed

[mookydooky]

I figured out a way to benchmark the two cores. On my BP6, when the 2 CPUs are running the same protein series, they tend to have almost identical times, usually off by about .01 minutes. I installed a new core on one CPU. I'll know sometime tomorrow whether there's any speed difference between the 2 cores. If the new core is substantially faster, it could mean an increase in output without having to upgrade anything.

[mookydooky]

It appears that the new core is substantially faster. Results are as follows:

Core V2.43 = 34.73 Min per frame
Core V2.41 = 38.35 Min per frame.

This works out to about a 9.4% advantage for the new core. I'm going to change the cores on all my machines as soon as I get a chance.

9.4% may not seem a lot, but if you calculate that as x100 minutes per WU, it adds up, especially if you have multiple machines. The difference for these two CPU's will work out to 362 minutes which is 6 hours!

[Lionfish]

What about core v2.44?

[mookydooky]

As soon as the Core V2.41 CPU is finished with it's WU (still a ways to go at 38min x 60), I will change that core to the 2.44 version to see if there's a difference in performance between the V2.43 and V2.44. Hopefully I'll have an answer by Monday.

[mookydooky]

Also, I don't know if any of you guys noticed, but there's a substantial size difference between the V2.41 and V2.43. V2.41 is about 600K whereas V2.43 is about 1.5M which makes it about 3 times bigger. I guess they added some stuff to make it do more per CPU cycle. If they changed something to increase the performance of the FPU, then there could be a much larger performance increase for K7 machines. ATM I don't have a dual K7 rig or 2 identical K7 machines so it's very difficult to gauge for me. Also all the K7 machines I have access to are in use for other stuff, so unlike the BP6, which is a dedicated folding box, I won't be able to obtain exact benchmarks. I'll figure something out though.

[mookydooky]

Originally posted by Lionfish
What about core v2.44?

Lionfish,

Is the V2.44 core working out for you? Any hiccups yet?

[Lionfish]

Running A OK

If I had problems you guyz would know

[SteveI]

I'm not sure if I have anything material to contribute but...
I just got my main machine back up. ATM it is only running 1500, but before @1542 it took about 9.9min for protA. I have one now and it seems to be going at about 9.5min. I don't know what the old core on it was (months old), but the core I am using now on it is v2.43. This is on an AthlonXP so I think Mooky's conclusion also applies to AMDs as well.

[mookydooky]

Just loaded Core V2.44 onto the other CPU. I should have an answer by the end of the day as to whether there's any speed difference between V2.44 and V2.43.

[mookydooky]

This may be a little premature, but these are the times I'm getting after 5 frames of V2.44

V2.43 = 34.06
V2.44 = 30.86

The V2.43 times is less than the last one because there were some background tasks going on the last time (windowsupdate) for about 20 minutes. The 34.06 should be a better measure. If these times hold up, it seems that the V2.44 is by far the best version, more than 20% faster than V2.41 and more than 10% faster than V2.43. I'll check the times again after it's done about 20-30 frames. I don't expect much of a change though.

[Allan]

WOW; this could really have some good inpact on my folding scores on the silly old laptop!!

[SteveI]

Is v2.44 a live version or is it a beta?

[mookydooky]

I'll have the final conclusions tomorrow night. I'm waiting for the second CPU to finish chewing up the WU it's currently working on so I can get a more exact figure. Also, I think the 2.44 version is a beta, but if it works better, who cares.

[mookydooky]

Ok, here are the final tallies after they have finished over 90 frames each.

V2.43 = 34.30
V2.44 = 34.19

It appears that there is not as much of a difference as I previously though. If you are not having any problems with the 2.43 core, you should stay with it. If you are still using the 2.41 core, you should definately switch since you will more than likely experience a 10% performance boost.

[S_Wilson]

I just went through looking at the fahlogs on my rigs and discovered that everyone of them is using the 2.44 core. I knew that my machines downloaded new cores automatically a week or so back, so maybe they are not beta anymore. The 2.44 works fine for me also.

[xylus]

There has been a lot of talk on other forums about the 2.44 core and most of it says that it is slower than the 2.43 core, though the 2.43 is a lot faster than the 2.41.

[mookydooky]

Originally posted by xylus
There has been a lot of talk on other forums about the 2.44 core and most of it says that it is slower than the 2.43 core, though the 2.43 is a lot faster than the 2.41.

I have both cores running on my one machine. Like I stated at the start of the thread, on the BP6, the two CPUs are never off by more than .01 minutes with the same core. Right now, one is running the 2.43 and the other is running 2.44. The 2.44 has a .11 lead on it right now. This was the same lead 3 days ago. Until someone can show me a better way to benchmark the speed of the cores, I will go along with the assumption that I'm correct.

[Allan]

I just put the 2.43 on the laptop. I hope it speeds it up a bit.