• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Dual Core vs Dual (seperate) Procs

R

RanceJustice

Supreme [H]ardness
2FA
Joined
Jun 9, 2003
Messages
7,064
Hey guys,

I've been wondering which would provide higher performance: one of the upcoming dual core procs, or 2 physical processors? Anyone have a hypothesis...

In a related question, I'm considering getting an AMD proc in my next rig. However, I've been rather happy with my Northwood's Hyperthreading capabilities. If I was to get a Dual AMD64 setup, would I get overall better performance in say...video encoding? I'm sure I'd get the advantage in single proc apps. Thanks
 
I'd say the dual cores and dual processors would give you the same benefits. Dual core means just putting two CPUs onto a processor, so you'd basically have a dual processor system.

As for benefits in terms of encoding: If the app is multithreaded or is otherwise able to utilize dual processors, then you'll notice a benefit. Intels still have a slight edge in encoding, but honestly, I haven't seen a dual Opteron system comparison against a regular Pentium 4 single proc system.
 
A dual Opteron should smoke a single P4. Hypertheading isn't like having 2 CPUs. It's more like having 1.5 or so. According to what I've read w/o hyperthreading a P4 is typically using about 1/3 of it's execution units at a time. Hyperthreading ups that to about 50%, hence the 1.5 approximation. Of course, you almost never get a 100% speed boost going from 1 to 2 procs, so the real number for the HT P4 is probably more like 1.3 or 1.4 times as fast as a P4 of the same clock speed w/ HT off.
The dual Opteron setups have also been beating pairs of HT Xeons in a lot of tests.
Now dual procs vs. dual core is another story. A dual core A64 on socket 939 would be kinda like having a pair of socket 754 chips w/ half the cache. I'd expect a dual core A64 could be quicker than a pair of single core Opterons w/ the same clock speed & same amount of cache. Sure you get half the mem bandwidth due to not having 2 mem controllers, but you toss the extra latency of the HT accessing mem on the other proc. Assuming the L2 cache is shared it'll eliminate having to maintain L2 cache coherency between two procs. Duallies have to do a lot more cache flushing than single procs since when one proc changes something, the other one has to throw out the cache line it's in if it's got it cached. The decreased latency from eliminating the HT link between procs & added efficiency in L2 cache handling just might be enough to push the dual core past a pair of single core Opterons at the same clock speed, despite the Opteron's advantages of double mem bandwidth and more total L2 cache.
 
Nice explanation and speculations Zandor. Yeah, I've seen dual opterons and dual xeons go at it, and the tests do lean more towards dual opterons.

On a somewhat related note, has anyone read the latest CPU magazine? They have an article about the Dallas Rock Festival and how they were using dual Opteron DAWs to record the 11-hour rock show (which featured Clapton, B.B. King, and others). They had an interview with head of audio at AMD and he explained how dual Opterons have an edge over Pentiums when you're doing these kind of live audio recordings, but I forgot his explanation of it. I'll have to get the issue out again.
 
These type of question all come down to bandwidth.
How well the two cores talk to each other.
How much bandwidth is there to the memory and is it shared.
How the cores talk to all the other chips/ slots etc.

Which is one reason why the Opteron's are at this moment better than the Xeon's.
Better memory bandwidth, its really noticably when you go 4 way.

Then again you could go for the best of both worlds and get a dual cpu, dual cored box when they come out.

Luck......... :D
u=Tigerbiten.gif
 
Tigerbiten said:
abbreviated....
Then again you could go for the best of both worlds and get a dual cpu, dual cored box when they come out.
Click to expand...

That's what I'm hoping to do. :D Oh, when that day comes...my wallet will be ready.

AMDs are reportedly working on a dual core Opteron as we speak, though it probably won't be out until same time next year.
 
Wow, thanks for the answers guys!

Hmm, will there be any chance of a dual proc, dual core A64 FX setup type in late 2004? Say november or december? I know it may take some time for the Opteron setup, but maybe the Prosumer version will be out earlier?

Hopefully this will be paired with a good mobo, with dual PCI-E slots.
 
Something i would love to see with dual core would be a type of load balencing between the two cores, so none multithreaded progs could use both cores. easier said than done, but that would rock
 
sigh, guess it will be awhile then. Well, I guess its dual AMD64 FX or Opteron64s in the fall, unless intel can come out with something better.
 
It is going to be awhile before any dual core proccessors are out. I would think that a dual core would be better than 2 seperate CPUs. But we I guess we have to wait for the benchmarks. I am also wondering how they expect to cool a dual core chip? Esp. Intel, imagine a dual core presscot... it could melt the motherboard.
 
You must log in or register to reply here.
Back
Top