**The LINPACK Thread**

[Puddle Jumper]

Re: **The Official TF LINPACK Thread**

Well I think you'll find people saying the same thing about LinX once we see some AMD numbers.

True, although the more benchmarks people claim are biased the less believable it becomes.

 

[zmatt]

Re: **The Official TF LINPACK Thread**

Ok, to alleviate the problems with the x64 guys I'm going to allow you to run it in 64bit for compatibility reasons. However take those numbers with a grain of salt as they will be inflated a bit over the 32bit ones.

Well I think you'll find people saying the same thing about LinX once we see some AMD numbers.

LINPACK is used on AMD, Intel, Cell and everything else. It's the main supercomputer benchmark. it's an industry standard. So if someone is complaining about fairness I'm going to call ricer foul and say they are sore looser.

Also Super PI is single threaded, 32bit only and very Intel biased. It hasn't been updated since '95 or so. It's a terrible benchmark. if LINPACK is good enough for IBM, NASA, and everyone else who has a super computer it's good enough for me.

 

[hikaricloud]

Re: **The Official TF LINPACK Thread**

I smell some overclocking tonight... >:]

 

[yzmxer608]

Re: **The Official TF LINPACK Thread**

So should we still use whatever problem and memory size that "All" gives us?

 

[septic]

Re: **The Official TF LINPACK Thread**

just run 2 tests. one with 12,000 and one with 15,000 ruffly. I couldn't run 15,500 myself as you saw in my failed test. I ran 12k and 14k at 3.0 and 3.3 succesfully and my gflops was always around 16-17.

 

[zmatt]

Re: **The Official TF LINPACK Thread**

just run 2 tests. one with 12,000 and one with 15,000 ruffly. I couldn't run 15,500 myself as you saw in my failed test. I ran 12k and 14k at 3.0 and 3.3 succesfully and my gflops was always around 16-17.

All determines the max problem size that your memory can hold. Setting a number outside of the parameter will give you errors. Stick to all. It isn't the same, but it does max out each computer's memory which means that we wont be under stressing anyone with 8 gigs of ram and 64bit.

 

[yzmxer608]

Re: **The Official TF LINPACK Thread**

That's true, I never thought about it that way (stresses everyone's system to it's max no matter what it is).

 

[wafflehammer]

Re: **The Official TF LINPACK Thread**

Yea, I think we should redo the list and let everyone run it in a way that would take advantage of their system. I mean it is a benchmark of their system so if you're scaling down the test to let everyone else run it, then you're not giving them room to show off.


So I say press the reset button and lets start fresh with no guidelines on the settings

 

[hikaricloud]

Re: **The Official TF LINPACK Thread**

Yea, I think we should redo the list and let everyone run it in a way that would take advantage of their system. I mean it is a benchmark of their system so if you're scaling down the test to let everyone else run it, then you're not giving them room to show off.


So I say press the reset button and lets start fresh with no guidelines on the settings

There wasn't really any guidelines, the memory was set to "all" in the first place...

 

[Hefemeister]

Re: **The Official TF LINPACK Thread**

You forgot to post your score.