• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Generic CPU Bench

anandtech said:
The other macro-ops fusion enhancement is that now 64-bit instructions can be fused together, whereas in the past only 32-bit instructions could be. It’s a slight performance improvement but 64-bit code could see a performance improvement on Nehalem
Fundamentally, on Core 2 is better executing 32 bit code than 64 bit code. The ONLY time 64-bit code will win over 32 bit code is when using shed loads of 64-bit data, which can be done "in one step" rather than multiple step under 32 bit.
 
Read the thread.
I dont find that a helpful, friendly or useful comment in any way.

The OP has been frequently updated. There is no request to use and report ONLY the x86 bench. The scoreboard needs to be ONE or the OTHER, or there needs to be TWO scoreboards. Clarity required in OP.

DONT make users have to read 156 posts for some buried information that you have at your fingertips.
 
I dont find that a helpful, friendly or useful comment in any way.

The OP has been frequently updated. There is no request to use one bench or the other. The scoreboard needs to be ONE or the OTHER, or there needs to be TWO scoreboards.

everyones moved over to the x86 version as it gives faster results. when i posted results in x64, i was told to use the x86 version.
 
Although it is a bit of work for btar, then AMD64 results should therefore be taken out of the table, and the OP should mention this requirement.
 
I dont find that a helpful, friendly or useful comment in any way.

The OP has been frequently updated. There is no request to use one bench or the other. The scoreboard needs to be ONE or the OTHER, or there needs to be TWO scoreboards.

Yet the problem and solution are both in this thread, so reading it is both helpful and useful. Would it be friendlier if I'd give you a lolly or icecream when you're reading it? Or does that only work for dentists?
 
Try not to be so grumpy in the morning. The problem is rudeness. The solution is courtesy and information.

You're right, they're both in the thread!

P.S. I'll call you on the ice-cream please.
 
Try not to be so grumpy in the morning. The problem is rudeness. The solution is courtesy and information.

You're right, they're both in the thread!

P.S. I'll call you on the ice-cream please.

grab me a cornetto on the way, plz.
 
Try not to be so grumpy in the morning. The problem is rudeness. The solution is courtesy and information.

You're right, they're both in the thread!

P.S. I'll call you on the ice-cream please.


Agreed, courtesy = read the thread. According to your logic we have the same discussion every page because people should not read the thread. If you take the courtesy to read up on a thread it won't get stuck in an endless loophole.

Also I do agree that bta could add the information to the op, that's a different matter though.
 
it doesnt matter if the scoreboards include x86 or x64, because it seems x86 is faster on every system. That means, anyone can run the x86 and compare to everyone else.

We would only need a seperate scoreboard, if x64 was faster, and therefore not everyone could run it.
 
Originally, the app was made such that it will run 64-bit in a 64-bit environment. So it's by your choice, you're forcing it to run 32-bit, why should I prescribe which app you run? As for mentioning in the scoreboards, it will make a clutter out of an already sumo 1st post. I can add a column on the first table mentioning which environment the thing was run on, but wouldn't the take the fun off ranking people based on the throughput, and the remarks? Those who secured high scores in 64-bit and left, would be owned by people who optionally chose to force 32-bit in a 64-bit environment. Your environment is part of your system, just as your processor. The app should behave in relation to your environment, so I wouldn't spoil the scoreboards with "he has the highest score, but he forced 32-bit on 64-bit". Let it be abstract. If someone is interested in how you did it, he'll be a nice student and go through the thread. If you benched on 64-bit and want an improved score, use the second app..you choose what to bench it on.

There are people who benched on 64-bit and left the thread, so I'm neither removing the original app, nor removing 64-bit entries.

Added a little note for newcomers to this thread.
 
Last edited:
Originally, the app was made such that it will run 64-bit in a 64-bit environment. So it's by your choice, you're forcing it to run 32-bit, why should I prescribe which app you run?

You don't have to prescribe, though some pointers on how to improve scores are welcomed by many.
 
Heres mine E8600 @ 5250Mhz On Modded Mach II Phasechange.

 
Last edited:
Yea AMD, ever upward
genericbench3.jpg

I switched memory to 1T
 
^^hey MKMOds, great scores dude, I just got my new ram, so time to play :)
 
looks like running forced 32 bit did a huge difference for me. My best at 3.4 GHz was about 23.5K. Now at 3.2GHz I got 28242 :)



 
best run till date

 
Man, I wish I had access to my phenom rig right now :(
 
where is it?

It's up in Traverse City at my parent's house. I'm moving and I left it up there because it's a bulky PC and the wifi on it was giving me crap. Right now I'm using the rig from my case mod.
 
sweet, broke 30K
30ksuka.jpg
 
Yeah... the server was actually running when I did this. Scores might be a little low :laugh:

Capture002.jpg
 
Does anyone actually know what this application is working with? Is it using addition, subtraction, multiplication, and/or division? Is it operating on floating-point decimals or integers? Are they 64-bit or 32-bit operations? Does it only support up to 4 cores?
 
Back
Top