C2D v X2 = Not as good as you think?

[Steves_pc]

Torque v Ponies.

Try a Yam TRX850 pretty good fun, maybe an acquired taste too though.

Back ot, i tend to run quite a bit all at once here too and i'm just about to stick a c2d together and i'm hoping your theory/findings are wrong!

As you've said the AMD's may slow a tad but dont really seem to stop and chug.

 

[fornowagain]

Erm, no.

Correct my ignorance here, but the IDE/SATA ports can go on either

I don't know the board. What I do know is that the Sata controller without AHCI drops back to a compatible slower IDE mode with Sata drives. That ain't going to help. I don't mean any disrespect to you, but RTFM comes to mind. I refrained from commenting, some emotive remarks here. But I'll say my piece. I came from an Opty [email protected] after 3 prior X2's. This is the first Intel I've owned in decades and honestly what your describing, it just sounds like something is wrong. I've never seen anything like it with mine.

 

[FatRakoon]

I don't know the board. What I do know is that the Sata controller without AHCI drops back to a compatible slower IDE mode with Sata drives. That ain't going to help. I don't mean any disrespect to you, but RTFM comes to mind. I refrained from commenting, some emotive remarks here. But I'll say my piece. I came from an Opty [email protected] after 3 prior X2's. This is the first Intel I've owned in decades and honestly what your describing, it just sounds like something is wrong. I've never seen anything like it with mine.

It does not say anything in the manual about me having to have ACHI mode on though?

That previous line also points out the fact that I have RTFM

Im not ignorant to trials though and I will HDTACH my setup with all the drives, set this ACHI thing to ... Erm, well... ACHI instead of IDE and re-do the HDTach tests and see what gives?

After all the Drive is SATA II and already its fairly quick at everything, so, who knows?

 

[FatRakoon]

Ok, I have gone into the BIOS, and set it to ACHI instead of IDE and in spite of trying to reinstall the drivers for it. I now have 6 of those IDE channels... 3 Primary and 3 secondary?

Am I correct here in thinking that the IDE ( Ok, the SATA ) drivers are not actually being installed correctly?

My list of ID Econtrolers now are :-

Intel(R) ICH8 2 Port Serial ATA Storage Controler - 2825
Intel(R) ICH8 4 Port Serial ATA Storage Controler - 2820
Primary IDE Channel
Primary IDE Channel
Primary IDE Channel
Secondary IDE Channel
Secondary IDE Channel
Secondary IDE Channel
Standard Dual Channel PCI IDE Controler

So, a bit too much I think there?

Also I was ( While Im here ) wondering if anyone else can verify the RAM figures that coming up for me???

Im currently playing about with the GEIL Ultra low latency stuff at the Moment, and I ran CPUZ to have a look at the settings and I have got

under memory:-
CAS# = 5.0
RAS to CAS Delay = 7 clocks
RAS Precharge = 7 Clocks
Cycle Time ( Tras ) = 20 clocks?

under SPD its telling me that the RAM is PC2-5300?? Week 39 Year 06

Anyway, I am currently at 415FSB x 7 giving me 2.9Ghz and I ran SuperPI and its at 22 Seconds?? - its a hell of a lot slower than the corsair @ 400FSB

That cant be right surely? I know the figures are pointing to a slow system, but come on! - not this much slower surely?

Im going to have to go back in now arent I?

Anyway, the Driver / Controler issue I think is my main concern for now... Its always done this however, but if anyone could very this on a DS3 please?

 

[marscay]

don't take this the wrong way fatrakoon but you are always starting threads relating to poor peformance or problems in one way or another with hardware.

c2d is faster than a64 that's just a fact, and once clocked up it's a lot faster.

i'd say there's every chance you don't something configured correctly.

 

[WJA96]

I've been following this thread and I'll be honest when I read the first post it hit a massive nerve. I am hugely into Folding@Home. I have a total of 6 machines folding for 4 different teams. All of these machines are overclocked E6600's and the performance is absolutely shocking from all of them. I was getting better performance out of my old Pentium 4 HT rigs.

I have never been a fan of AMD. I don't play games much and I like the 'stability' of Intel. I hang around with server guys and they pretty much like Xeons, Xeons and more Xeons although Opterons are starting to feature in conversations now and again. The point is - I could never be accused of being an AMD fan-boy.

Anyway, I borrowed an AMD 4800 x2 system from a colleague (who was absolutely delighted to be taking home my EVGA 680i/8800GTX for the weekend) and I downloaded a work unit. It's an 1167 Gromacs (494 pointer) and I set the 2 machines running the same work unit of both cores. A waste, yes, but this is science!

Now they are not identical because of the motherboards but each machine is set up roughly the same.

2Gb Patriot DDR2 PC5300 RAM at stock speeds and timings
A fresh install of XP Pro SP2 with all updates applied
AVAST Antivirus
A 36Gb Raptor 8Mb Cache drive formatted to NTFS
A PCI graphics card (one is a 5200, the other a 6200, but it shouldn't matter as both are 128Mb RAM and not Turbocache)

Both machines are running at as close to their normal speeds as the trial conditions will allow; the 4800 is not overclocked because my colleague doesn't do that sort of thing. The Core2Duo is overclocked to 3.0Ghz as that's the top I can give it in a DS4 without overclocking the RAM. NOrmally it would run closer to 3.4GHz.

On the motherboard front we're looking at a DS4 vs. ASUS M2R32

Now, I could be spectacularly unlucky in my choice of work units but the AMD has absolutely spanked the Intel in this test. The AMD finished last night and did almost 300ppd per core while the Intel is struggling to finish this before lunchtime at 208ppd per core.

Now, I'm going to let my colleague keep the flashy Core2Duo system for a week just to see if the AMD will earn more points for my teams than the Intel, but I am starting to genuinely think that for Folding@Home at least, FatRakoon is absolutely bang on. Core2Duo is not faster than AMD X2.

 

[PikeyPriest]

stuff

I`m a bit of a folding newb, but i thought some units are meant to be faster/slower depending on what processor you were running? I `m sure i have read that depending on the core being used, some run faster or slower on either AMD or intel. Does anyone know any more about this? Or was it just speculation?

 

[DanF]

Yeah from the first post it does seem to indicate a problem with FaH and the C2D system. Seems a bit strange that it's still loading the cpu when other programs are trying to take priority.

I guess others is that the 6600 might preform better than the 6300 in mass multi-tasking due to the extra L2 cache (it needs more cache than an equivalent AMD system due to the architecture and off-chip memory controller). Also aren't a lot of opteron systems faster due to the memory bottleneck with Intel systems and multi-cores/cpu's in server environments which is essentially mass multitasking.

 

[mrthingyx]

Could it be something to do with the memory latency suffered by both CPUs? The on-board controller in the AMDs has been shown to make up for a lot of ground in sorting through workload, whereas the C2D has to suffer with the access limitations of the FSB?

No idea, personally, although that an X2 may bear a massive and diverse workload better than a C2D doesn't surprise me that much.

However, the level of petty name-calling and bickering in this thread does. It is quite sad to see it, regardless of how much people have invested in a particular system.

 

[DanF]

@FatRakoon - Could you try re-testing with FaH turned off? And have Task Manager running to make sure your not hitting the page file as this would kill performance especially where most of you HD's are already being accessed.

 

[WJA96]

@FatRakoon - Could you try re-testing with FaH turned off? And have Task Manager running to make sure your not hitting the page file as this would kill performance especially where most of you HD's are already being accessed.

I was thinking that too - maybe F@H is the problem. It is, after all, written by a bunch of students and they admit the testing is quite light. Pretty much everyone is running a Beta client.

 

[WJA96]

I`m a bit of a folding newb, but i thought some units are meant to be faster/slower depending on what processor you were running? I `m sure i have read that depending on the core being used, some run faster or slower on either AMD or intel. Does anyone know any more about this? Or was it just speculation?

Yes, you are absolutely correct, which is why I'm keeping the system for a week at least (maybe over Christmas and new year too) to see if my numbers go up compared to the Intel systems. What I do know is that I replaced 10 P4 HT units (single core, hyperthreaded) with 6 Core2Duo dual cores and my folding scores are about 30% of what they were. I could have saved myself about £1800 and just kept the P4's to be frank.

 

[FatRakoon]

Sure things guys. I will kill off the FAH clients on them and then have another play about later on.

I read about FAH being strong on Intels for X Projects and good on AMD for Y Projects, so maybe...?

Me starting threads relating to poor performance?

Yes, I am guilty as charged on that one.

I will be honest here, but when I moved from a Barton 3200 to my winchester 3200, I found very little difference in performance, while everyone was talkign about how its twice as fast blah blah, when it quite simply is not. I was simply looking for this so called 200% boost that never of course came.

I accepted it and after a while I realised that it was quicker but only at this or that but not the other.

When I upgraded the Winchester to a Venice, because people were saying it was better cos of this or that, and I also realised that the winchester was quicker than the venice, so what are all these people on about? - some extra commands that nothing I have will probably ever use, thats what.

When I went to the X2 3800 I was expecting double the speed, but I simply never got it... In fact, the winchester clocks way more than the 3800 does and so in fact I think the Winchester is a better CPU than the x2-3800 for most things, and to be honest, even today I only hve one program that I use regularly that can useboth cores and thats DVD Santa... Everythign else I dont really need 2 cores for.. Not really. I set most of my jobs up and they are all ready for me to click on go, and when I do, I often just walk away anyway, so whether it takes 1 hour on the winchester or half an hour on the 3800, I really dont care too much.

It was the same when I went for the Opteron assuming that that extra cache would let me work the system harder... It has not... Actually I went for the opteron expecting it to be dual core to be honest, but thats me all over... It looked cheap and I didnt read past the price.

No, the simple truth is that I have upgraded either a tiny ammount expecting it to be more than it is, or even worse, I have downgraded in some ways. I sometimes think that maybe its me, and so yes, I post and sometimes I get answers, and a lot of the time I get flames, but that I really dont care about... Not anymore.

 

[mrthingyx]

Hmm... just looking at memory bandwidth/latency and the AMDs absolutely destroy the Conroe chips (9kMB/sec for the FX-62 vs 5.6KMB/sec for hte X6800).

Which may account for better extreme multitasking if available memory is constant.

 

[Cyber-Mav]

hmmm, wonder if the conroe architecture is same as the dothan cpu i have, that hates multitasking but works real quick if its given 1 task to do at a time.

im guessing the on die memory controller is probably the cause of amd's quick time slicing between apps. not sure though.

to be honest no one else i have ever known or even seen has run so many tasks on the pc at one.

for me the most i have is winamp and ie7 running at the same time with a msn window. so thats not much at all. yea sometimes i burn a dvd too but its not as demanding as you have your system run stuff.

need to find out more by running more tests cuz every test iv seen the c2d beats amd BUT those tests are run singularly as in only the test is running and nothing else. so your kind of test environment is to the extreme and never before done.

wonder if there are more people out there that have the 2 hardwares to test this out further.

 

[WJA96]

I'm just trying the F@H thing but at the moment it's AMD 1: Intel 0

I think it's perfectly feasible Intel has designed a product to be good at benchmarks. I can't believe they don't look at this before they launch new products.

 

[clv101]

No, the simple truth is that I have upgraded either a tiny ammount expecting it to be more than it is, or even worse, I have downgraded in some ways. I sometimes think that maybe its me, and so yes, I post and sometimes I get answers, and a lot of the time I get flames, but that I really dont care about... Not anymore.

And the moral of the story? Don’t upgrade so much! My CPU history has been:

386sx25MHz
Pentium 60 @66MHz
Pentium 150 @187.5MHz
Celeron 300A @464MHz
Athlon AXIA 1GHz @1.4GHz
Pentium4 2.8 @3.2GHz

Each upgrade has been significant, allowing me to do things I just couldn’t do before and always great value. I’m now thinking about an E6600 @3.6GHz as a decent upgrade from the P4.

 

[22BUK]

I think it's perfectly feasible Intel has designed a product to be good at benchmarks.

Why on earth would they do that? Who pays attention to benchmarks anyway?

 

[DanF]

I'm just trying the F@H thing but at the moment it's AMD 1: Intel 0

I think it's perfectly feasible Intel has designed a product to be good at benchmarks. I can't believe they don't look at this before they launch new products.

heh,

Well more likely that FaH hasn't been optimized for the C2D architecture yet. The C2D does great in most multi-threaded apps like Photoshop filters etc.

 

[mrthingyx]

hmmm, wonder if the conroe architecture is same as the dothan cpu i have, that hates multitasking but works real quick if its given 1 task to do at a time.

im guessing the on die memory controller is probably the cause of amd's quick time slicing between apps. not sure though.

Yup - the Dothan was the base structure for the C2D... both chips are rooted firmly in P3 architecture. But then the Athlon 64s use the Barton FPU, so go figure...