DPAD 4.34 performance

[MechEng]

Have anyone else but me tried to bench version 3.34 ?
The results shown are averaged from running the clients for +12 hours each.

DPAD 4.33a: Average: 282,45kpts/sec

DPAD 4.34: Average: 284,79kpts/sec

These benches show an increase of approximately 0,8%

While it is not much, it is something...

 

[Orange Kid]

Something is good

Over the course of a week, that should yield another 66k mpts or so for the project from our Team alone.

As they say elsewhere on these forums ...more,more,more

 

[Zbox]

I only rolled a couple of my systems over to 4.34, and no testing done yet...

Is that a HT cpu you're running, Mech?

 

[MechEng]

Originally posted by: Zbox
I only rolled a couple of my systems over to 4.34, and no testing done yet... Is that a HT cpu you're running, Mech?

Yes, it's a P4 2,8@3,16

Details can be found in my sig.

 

[MechEng]

Originally posted by: MechEng
Have anyone else but me tried to bench version 4.34 ? The results shown are averaged from running the clients for +12 hours each. DPAD 4.33a: Average: 282,45kpts/sec DPAD 4.34: Average: 284,79kpts/sec These benches show an increase of approximately 0,8% While it is not much, it is something...

Update:

DPAD 4.34: Average: 285,95kpts/sec

Increase of approximately 1,24% over 4.33a

Anyone else get these improvements, or....

 

[BlackMountainCow]

24h bench gave me 0,87% advantage ... but I guess this is within the limits of benchmarking inaccuracy

 

[MechEng]

Originally posted by: BlackMountainCow
24h bench gave me 0,87% advantage ... but I guess this is within the limits of benchmarking inaccuracy

I would say that would be the case if you only run it for a few hours, but for every hour that you run it, the accuracy becomes better I suppose. If you run it for five minutes, and you browse the web meanwhile, then the overall penalty is larger, than if you had run the bench for 24hours, and still only browsed the web for five minutes.

 

[GtPrOjEcTX]

in contrast, with 4.33a, running two threads and with 4.34 running two threads, I notice more of a hit in performance with everyday use of the computer.

 

[BlackMountainCow]

Originally posted by: GtPrOjEcTX
in contrast, with 4.33a, running two threads and with 4.34 running two threads, I notice more of a hit in performance with everyday use of the computer.

Stephen said that he tweaked the dual or HT support in this release a bit ...