• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

SuperPi...

Tee Hee Johnson

Tee Hee Johnson

Tee Hee Johnson

Soldato
Joined
30 May 2007
Posts
4,970
Location
Glasgow, Scotland
Hiyahh! :D

I was sorting out mums laptop today, and decided to run superpi mod 1.5 xs on it :)

Its a C2D meron (or something) 1.7Ghz (T5300 i think) and it done 1M in 34.602 seconds!

The system in my sig done it in 38.190!

Does this show the sheer upperhand Intel have at the moment?! Didnt think things were too different! Oh and i know this isnt real life game play, but surely raw power is good?

Sorry if this is a pointless post...!

Craig!
 
Intel Core2's perform very favorably at SuperPI, you're very likely to get an Intel and an AMD which perform very similarly in the real world but the SuperPI test on the Intel will be much faster - It's a purely synthetic benchmark. Not saying that Core2's are bad :p
 
craig_kennedy83 said:
Hiyahh! :D

I was sorting out mums laptop today, and decided to run superpi mod 1.5 xs on it :)

Its a C2D meron (or something) 1.7Ghz (T5300 i think) and it done 1M in 34.602 seconds!

The system in my sig done it in 38.190!

Does this show the sheer upperhand Intel have at the moment?! Didnt think things were too different! Oh and i know this isnt real life game play, but surely raw power is good?

Sorry if this is a pointless post...!

Craig!
Click to expand...

Are you sure your rig is that slow? I swear my old 939 system (x2 3800 & 2GB RAM) did 1M in about 40 seconds at stock?
 
The AMD64 rigs are much slower than their C2D equivalents.

The benchmark used to be a very general indicator of system performance, but it doesn't seem to be quite as reflective in the current processor environment.

Now, this could be sour grapes on AMDroids' part but something else bears thinking about... a Phenom clocked to 3.3GHz actually benches slightly higher than a /q660, and yet does Superpi in 23secs as opposed to around 12 for the Q6600.
 
Super Pi is not a reflection on performance.

Below a is quite a fast time yes, 10secs faster than a Phenom at 3.3ghz. but will that be reflected in 'real' performance. I cant' see it personally. I'd expect to see the Phenom @ 3.3ghz beating my chip in most tests.

4000pi.jpg
 
ManCuBuS said:
snipQUOTE]

I think you misunderestimassumed my post (either that or I have yours...), Mancubus.

SuperPi was a benchmark that - prior to Core2Duo and latter AMD64 chips - showed reasonable (nothing more) correlation with average system performance in other benchmarks and was widely used to compare "overall" performance for P4 vs. XP.

Now, however, CPUs are somewhat different in their architecture and so actual performance no longer correlates well with SuperPi times. That is not to say that SuperPi ever was a valid tool for comparison - only that it was used that way for a few years.

Now, just as SuperPi is not the be-all and end-all of performance assessment, the number of cores is not the prime driver of performance either. Some applications respond better to pure MHz and others to cache sizes. The advent of multiple processors has changed the benchmarking landscape in that different tools have to be used to assess different aspects of CPU performance, i.e. using a single-threaded game to assess an octo-core system.

Even then we have the 'useability' of a system. A few people on this forum have posted colloquial comparisons of their AMD and Intel systems, noting that the AMD systems deal with multiple loads more effectively and that games run more smoothly that their equivalent Intel hardware. These observations cannot be explicitly validated by benchmark performance and explanations must be inferred from available data.

This is just an example of how one single benchmark - or even a suite of benchmarks - is still not wholly indicative of a system's overall ability, unless in those specific circumstances.
Click to expand...
 
I suppose its 'my bad' i dunno why, but i had it pegged as a kinda good overall perfomance tool... excluding graphics obviously!

Dunno why i thought that... musta been these forums, as thats where i heard of superpi :P
 
No need to apologise. It was 'accepted truth' a couple of years ago for the reasons I've posted.

Prime 95 is also lauded as the ultimate stress test when in fact it only tests half a PC (anywhere between the FPU, cache and memory): I'd be a reasonably wealthy man if I had a tenner each time somebody posted on these forums asking why - after running Prime95/Orthos for over eight hours (as is the commonly held dogma) - their computer fell over when opening Firefox or playing a game.
 
mrthingyx said:
No need to apologise. It was 'accepted truth' a couple of years ago for the reasons I've posted.

Prime 95 is also lauded as the ultimate stress test when in fact it only tests half a PC (anywhere between the FPU, cache and memory): I'd be a reasonably wealthy man if I had a tenner each time somebody posted on these forums asking why - after running Prime95/Orthos for over eight hours (as is the commonly held dogma) - their computer fell over when opening Firefox or playing a game.
Click to expand...

What would be a good substitute/addition to Prime/Orthos for stress testing the rest of the PC?
 
Super PI is a good way of testing how well arithmetic implementation of fft algorithms and quadratically convergent algorithms are handled by a CPU. these methods of computation are commonly used by programs to solve polynomial equations, different processors will have differing architectures that change the way they calculate specific sums. Benchmark tool that attempted to mimic real world usage workloads will give you a more reliable way of comparing processors.
 
craig_kennedy83 said:
To be honest there seems like no one application!

Think a good mix of Ortho, 3D Mark, Games etc seems to be it! :)
Click to expand...

I remember seeing a power chart of the Phenom when switching between a quad session of Orthos and UT3 and it was easy to see that the game drew almost 33% more power than Orthos.

Sure, it's involving the GPU, but there's one element where all the power comes from...

The only way to make sure your computer is completely stable (in all senses of the word) is to run it through the mill of the worst situations it could come up against.

- Orthos blend for 12hrs
- 3Dmark06 / Vantage for 8hrs on loop
- Encode a HD video at its highest bitrate
- Toast for 8hrs.

Toast is a funny little app: if you think your CPU is loaded running Orthos, Toast will really make it sweat.

That is what I'd consider a reasonable stability test for an overclocked PC. It sounds severe, but all components of the PC have to be stressed individually and together to see what it will do.
 
Thanks for the advice, currently trying out Toast, and running two instances as it seems to be single threaded.

I've also seen OCCT recommended elsewhere so I'm going to give that a run as well. My temperatures don't seem to get very high with Toast mind, almost 10C lower than when running Orthos, although my PC seems to be less responsive, even when setting it at low priority.
 
Toast is a very (comparatively) old test: it basically loops thirteen instructions through the CPU to see how much heat it can throw out. However, if it doesn't seem to be cranking up the temps for your system, try something else. I use it because I'm still on an Athlon XP... :D

But yes - as with Prime95, two instances have to be generated (or four) to fully load a multicore CPU.
 
You must log in or register to reply here.
Back
Top Bottom