I was just browsing throught the stats on vairous peoples computers and I have noticed the number of PC's that are faster than mine for FPU's and Measured integer speeds. This is to be expected though, but I noticed the following:

I have a Athlon 64 3500 and BIONIC rates my pc as the following:

Measured floating point speed 1869.85 million ops/sec
Measured integer speed 3769.7 million ops/sec


I looked at a different user and he had the following:

Authentic Athlon (not xp or 64)

Measured floating point speed 1914.96 million ops/sec
Measured integer speed 3253.12 million ops/sec


I fail to see even if the guys PC is overclocked it should be no way near the speed of the A64 3500. Can someone explain this for me?

Thanks

---

Perhaps this guy is using an optimized CC.

Happy crunching,

---

Perhaps this guy is using an optimized CC.

Happy crunching,

Even if it was optimized surely you could never get the CPU to perform as much as his states? I have another Seti client and it is a authentic AMD, its stats are:

MFPS 1116.08 million ops/sec
MIS 2006.87 million ops/sec


Thats a 1.33Ghz AMD the only one faster than that is the 1.4

---

I'm using the optimized CC 4.68 from boinc.truxoft.com along with the optimized seti client 4.11 from TMR.

My AMD 64 3000+ (Winchester) gets these benchmarks:
Measured floating point speed 1824.35 million ops/sec
Measured integer speed 6630.57 million ops/sec

Andy

---

Want to know your pending credit?


The biggest bug is sitting 10 inch in front of the screen.

Maybe it's your stats that are misleading? I have a 2400+ AMD using the non optimized client that benchmarks better:

hostid=1136801

Measured floating point speed 1872.21 million ops/sec
Measured integer speed 4548.22 million ops/sec

Could well an optimised boinc cc. I have compiled my own under Linux and got a 55% increase from the standard client. If that guy is also overclocking as well as optimised as well as making sure the PC is running nothing else when benchmarks happen then its feasible I guess.

---

To be honest, your benchmarks do look a little low. My AMD 64 3200 Clawhammer is nearly identical to yours under benchmarking, using the stock 4.45 CC and no overclocking.

Don't forget benchmarks are by their very nature synthetic, and can be affected by many things. An overclocked PC running an optimised client can BENCHMARK quicker than your 3500, even if in real life it is slower.

---

My athlon 64, 3200, just overclocked 7% or so, (as much as I can get yet, and still pass prime95 all day), is here:

http://setiathome.berkeley.edu/show_host_detail.php?hostid=1152608

Running the sse2 boinc core client, and a seti client, I get benchmarks of:

Measured floating point speed 2465.33 million ops/sec
Measured integer speed 3795.79 million ops/sec

I suggest you try out the optimized clients, and nearly double your actual throughput. All from here:

http://www.marisan.nl/seti/index.htm

Some folks are finding that the previous version 7, of the sse2 seti client, actually runs faster, on their systems. There is a reference WU you can run, and report here:

http://setiathome.berkeley.edu/forum_thread.php?id=17474

Let us know what you find, ok?

Don

---

Even if it was optimized surely you could never get the CPU to perform as much as his states?

Nope. The inflated benchmarks are fairly useless. They bear no real relation to what your machine is capable of.

By the way, the athlon xp mobile often shows up as just, Authentic Athlon, and they are favored by overclockers, since their core ratio is unlocked.

I had one running about 2.4 GHz, and it benched a bit higher than that even.

But in that case, the throughput wasn't there, since it did not have sse2, so can't use a really great optimized client.

Don

---

AMD mobile Athlon64 3700+

laptop, no overclocking

8/4/2005 9:11:51 PM|| 2217 double precision MIPS (Whetstone) per CPU
8/4/2005 9:11:51 PM|| 4097 integer MIPS (Dhrystone) per CPU