Floating Point or Integer?

Dex
Dex
Joined: 21 Sep 06
Posts: 646
Credit: 4,381
RAC: 0
Topic 192277

Pease excuse my crudeness on this request of info. But, I am going to be purchasing a new computer soon, and have been reviewing and contemplating CPU speeds, etc... I suppose, I could just scan over a few different computer summary reports, and get my answers, but, I will ask the pro's... I noticed that the AMD Sempron Proccessors produce higher Integer Calculations per second, and Intel Celeron proccessors, produce more Floating Point Calculations per second... Which is going to make the biggest speed difference with E@H WU's? Or does the combined usually meet somewhere in between? Thanks in advance...

d3xt3r.net

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,584
Credit: 310,507,331
RAC: 91,243

Floating Point or Integer?

Quote:
Pease excuse my crudeness on this request of info. But, I am going to be purchasing a new computer soon, and have been reviewing and contemplating CPU speeds, etc... I suppose, I could just scan over a few different computer summary reports, and get my answers, but, I will ask the pro's... I noticed that the AMD Sempron Proccessors produce higher Integer Calculations per second, and Intel Celeron proccessors, produce more Floating Point Calculations per second... Which is going to make the biggest speed difference with E@H WU's? Or does the combined usually meet somewhere in between? Thanks in advance...


I think FPU performance is vastly more important - there's a bucket of floating point arithmetic to do. Akos Fekete's optimisations last year, now incorporated as standard, was in the floating point area of trigonometric functions from my recollection.

Cheers, Mike.

I have made this letter longer than usual because I lack the time to make it shorter ...

... and my other CPU is a Ryzen 5950X :-) Blaise Pascal

Dex
Dex
Joined: 21 Sep 06
Posts: 646
Credit: 4,381
RAC: 0

RE: RE: Pease excuse my

Message 58971 in response to message 58970

Quote:
Quote:
Pease excuse my crudeness on this request of info. But, I am going to be purchasing a new computer soon, and have been reviewing and contemplating CPU speeds, etc... I suppose, I could just scan over a few different computer summary reports, and get my answers, but, I will ask the pro's... I noticed that the AMD Sempron Proccessors produce higher Integer Calculations per second, and Intel Celeron proccessors, produce more Floating Point Calculations per second... Which is going to make the biggest speed difference with E@H WU's? Or does the combined usually meet somewhere in between? Thanks in advance...

I think FPU performance is vastly more important - there's a bucket of floating point arithmetic to do. Akos Fekete's optimisations last year, now incorporated as standard, was in the floating point area of trigonometric functions from my recollection.

Cheers, Mike.

I was somewhat hoping that to be the answer, considering, I like to lean towards AMD. THANKS!

d3xt3r.net

Annika
Annika
Joined: 8 Aug 06
Posts: 720
Credit: 494,410
RAC: 0

Well dunno about float or

Well dunno about float or integer, but Einstein is at the moment said to be an AMD project... meaning that an AMD CPU of comparable speed will always beat an Intel, at least that's what I've heard and experience seems to confirm it. It's different with other projects of course... if you want to do SETI, do by every means get an Intel (preferably one with a large cache, such as for example the Dothan mobile CPUS... but that doesn't apply to Celerons so it's probably less interesting for you), whereas CPDN is slightly more suitable for AMDs aswell (though I think the difference is smaller there). But if you're going on 100% Einstein I'd advise you to get an AMD CPU.

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,584
Credit: 310,507,331
RAC: 91,243

RE: Well dunno about float

Message 58973 in response to message 58972

Quote:
Well dunno about float or integer, but Einstein is at the moment said to be an AMD project... meaning that an AMD CPU of comparable speed will always beat an Intel, at least that's what I've heard and experience seems to confirm it. It's different with other projects of course... if you want to do SETI, do by every means get an Intel (preferably one with a large cache, such as for example the Dothan mobile CPUS... but that doesn't apply to Celerons so it's probably less interesting for you), whereas CPDN is slightly more suitable for AMDs aswell (though I think the difference is smaller there). But if you're going on 100% Einstein I'd advise you to get an AMD CPU.


Yeah, I remember a 'consensus' opinion here that AMD burned FP much better than Intel.

Cheers, Mike.

I have made this letter longer than usual because I lack the time to make it shorter ...

... and my other CPU is a Ryzen 5950X :-) Blaise Pascal

Udo
Udo
Joined: 19 May 05
Posts: 203
Credit: 8,945,570
RAC: 0

RE: Well dunno about float

Message 58974 in response to message 58972

Quote:
Well dunno about float or integer, but Einstein is at the moment said to be an AMD project... meaning that an AMD CPU of comparable speed will always beat an Intel, at least that's what I've heard and experience seems to confirm it. It's different with other projects of course... if you want to do SETI, do by every means get an Intel (preferably one with a large cache, such as for example the Dothan mobile CPUS... but that doesn't apply to Celerons so it's probably less interesting for you), whereas CPDN is slightly more suitable for AMDs aswell (though I think the difference is smaller there). But if you're going on 100% Einstein I'd advise you to get an AMD CPU.

My AMD Athlon 1700+ at home finishes short WUs in approx. 3600 sec.
My Intel P4 2000 in the office finishes short WUs in approx. 4700 sec.

Udo

Wurgl (speak^Wcrunching for Special: Off-Topic)
Wurgl (speak^Wc...
Joined: 11 Feb 05
Posts: 321
Credit: 140,550,008
RAC: 0

RE: My AMD Athlon 1700+ at

Message 58975 in response to message 58974

Quote:

My AMD Athlon 1700+ at home finishes short WUs in approx. 3600 sec.
My Intel P4 2000 in the office finishes short WUs in approx. 4700 sec.

Please compare 'new' CPUs. Especially für Intel the P4 is not so good as the new C2D.

Apple C2D 2.33 GHz: short WU in 1680-1850 sec, long in 16200-17400
Monocore Opteron 244 on Linux 1.8GHz: short 2780-2960, long 24900-26300

If I transform the speed of the opteron to 2.33 GHz I get 2150-2290 and 19240-20300.

So when you compare the same speed, Intel is currently the winner. You need an 3.0GHz Opteron to beat a 2.33GHz C2D. But note: My Opteron is a relatively old one, bought in May 2005. Before C2D AMD was faster than the P4-Netburst.

Dex
Dex
Joined: 21 Sep 06
Posts: 646
Credit: 4,381
RAC: 0

Well, I am looking at getting

Well, I am looking at getting a Acer laptop with a mobile AMD Sempron 1.8GHZ 3000+

A laptop on my team, a Compaq Presario Mobile AMD Sempron 1.8GHZ 3000+ does a short WU in between 2700 to 2900 secs. So, I think that may be the CPU I go with.

I am going to search google.com on this site, to see if I can find a Intel Celeron 2.6 GHZ, which is the other proccessor I am thinking about. To compare times.

EDIT: Well after looking. I guess the Celeron I was interested in was a 1.46GHZ. Which is less fast. Unless I want to go up to a Mobile Turion 64 x2, I think the Mobile Sempron is the best option. Each proccessor on a Turion does a small WU in about 3600 seconds but, if you count in both proccessors working at the same time, it is fast. But, I think the Mobile Sempron is what I am going to go with.

d3xt3r.net

Desti
Desti
Joined: 20 Aug 05
Posts: 117
Credit: 23,762,214
RAC: 0

RE: Well, I am looking at

Message 58977 in response to message 58976

Quote:

Well, I am looking at getting a Acer laptop with a mobile AMD Sempron 1.8GHZ 3000+

A laptop on my team, a Compaq Presario Mobile AMD Sempron 1.8GHZ 3000+ does a short WU in between 2700 to 2900 secs. So, I think that may be the CPU I go with.

I am going to search google.com on this site, to see if I can find a Intel Celeron 2.6 GHZ, which is the other proccessor I am thinking about. To compare times.

EDIT: Well after looking. I guess the Celeron I was interested in was a 1.46GHZ. Which is less fast. Unless I want to go up to a Mobile Turion 64 x2, I think the Mobile Sempron is the best option. Each proccessor on a Turion does a small WU in about 3600 seconds but, if you count in both proccessors working at the same time, it is fast. But, I think the Mobile Sempron is what I am going to go with.

The desktop Celeron and the mobile Celeron M are two completly different processors. The desktop celeron is based an the Pentium 4 Netburst technologie, while the Celeron M is based on the Pentium M (Core 1) technologie.

Annika
Annika
Joined: 8 Aug 06
Posts: 720
Credit: 494,410
RAC: 0

Yep, we just discussed that

Yep, we just discussed that in another thread. A Celeron M would probably, in newer PCs, be a Dothan core (or maybe still the older Banias, which was already quite nice) and that is related to the Core- based on the old P3 architecture (Coppermine and similar) iirc and also related to the new Core 2 Duos, whereas the P4 is a completely different architecture. I think Core CPUs are way more efficient than Netburst if I see what my brother's Dothan (Pentium M sold as Centrino) can do in the benchmarks at only 1.7 GHz which sounds lame but is easily fast enough for gaming and stuff...

Svenie25
Svenie25
Joined: 21 Mar 05
Posts: 139
Credit: 2,436,862
RAC: 0

My C2D E6300@2,8 GHz needs

My C2D E6300@2,8 GHz needs around 1200 seconds for a short und 12800 seconds for a long WU. I think this is respectable for an Intel. ;)

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.