LHC@home

I looked at the results for the recent wu's I had. In all cases, machines equipped with AMD processors returned the units in substantially shorter times, often as much as 50% of the time.

This result for example. The Intel chipped machines both returned figures 15-16,000 seconds, the two AMD machines, ~7,000.

Another. Different figures but same trend.

Any comments anyone?

---

Wave upon wave of demented avengers march cheerfully out of obscurity into the dream.

One machine (yours) is an HT machine so in reality divide your cpu time by 2 to get an 'equivalent' AMD time. The other INTEL machine is a 2.4GHz chip so is slow in any case. The variation in claimed credit is fairly typical.

Live long and crunch!

---

Paul
(S@H1 8888)
BOINC/SAH BETA

> One machine (yours) is an HT machine so in reality divide your cpu time by 2
> to get an 'equivalent' AMD time. The other INTEL machine is a 2.4GHz chip so
> is slow in any case. The variation in claimed credit is fairly typical.
>
> Live long and crunch!
>
>
Yeah, that's a HT issue... :-)

---

Ignore my machine, in the first example I gave, 35696 is a 2600 Athlon and 22101 a 2400 Intel yet the Athlon cranked the wu in 7,755 seconds and the Intel, (non HT), took 15,953. That is a very big difference, less then half the time. Why is that?

---

Wave upon wave of demented avengers march cheerfully out of obscurity into the dream.

> That is a very big difference, less then half the time. Why is that?

Per clock cycle, AthlonXP & Athlon64 machines perform more work. The answer is far more complicated than that but the easist to go with. :)

---

Some of these differences are due to the differing development paths taken by Intel and AMD with their successive generations of processors. Intel have optimised their P4 around multimedia applications, games, et al, at the expense of the FPU. A P3 FPU would be faster than a P$ unit at the same clock frequency. AMD have concentrated their efforts in other ways, so their FPU is significantly faster than Intel's. On a like for like basis the AMD chip will outperform an Intel chip on floating point operations.

---

Gaspode the UnDressed
http://www.littlevale.co.uk

Must be project dependent as this S@H result has the longest crunch times for the AMD's and the Intels performing best.

---

Wave upon wave of demented avengers march cheerfully out of obscurity into the dream.

> Must be project dependent as <a> href="http://setiweb.ssl.berkeley.edu/workunit.php?wuid=17143831">this S@H
> result[/url] has the longest crunch times for the AMD's and the Intels
> performing best.
>
The P4s make up for the lack in the FPU with a larger cache. The seti app is small enough to fit into the cache most of the time so the P4s can do them quicker since there are fewer calls to main memory. The complier used and the options used when compiling can also make a difference.

---

BOINC WIKI

BOINCing since 2002/12/8