Cache memory bandwidth

Toggle sidebar Toggle sidebar
B

bigbadwolf

Distinguished
Feb 17, 2006
87
0
18,630
I was running a memory test the other day, the one at the beginning of SUSE linux install, and it stated that the L1 Cache was - 16,476MBps, and L2 Cache was - 3,925MBps.

Is this normal? What Frequency do L1 and L2 cache run at?

Also it stated my DDR400 is only 1,602MBps, you and I both know it is supposed to be around 3200,MBps. Maybe the memory test has measured these speeds wrongly? Because I would have thought that all these speeds are suposed to be faster, especially the L2 Cache.

(AMD ATHLON 64 3200 VENICE, ASUS A8N-SLI DELUXE)
 
I was running a memory test the other day, the one at the beginning of SUSE linux install, and it stated that the L1 Cache was - 16,476MBps, and L2 Cache was - 3,925MBps.

Is this normal? What Frequency do L1 and L2 cache run at?

Also it stated my DDR400 is only 1,602MBps, you and I both know it is supposed to be around 3200,MBps. Maybe the memory test has measured these speeds wrongly? Because I would have thought that all these speeds are suposed to be faster, especially the L2 Cache.

(AMD ATHLON 64 3200 VENICE, ASUS A8N-SLI DELUXE)
Click to expand...

Due to cache misses and latencies, your L1 Cache and L2 Cache will be different by alot, and 16GB/s for L1 Cache is good for the 3200+, I believe my 3700+ gets about 18GB/s. Since you have socket 939, you should be getting at least 4.5GB/s, even with valueRAM. A friend of mine has Kingston Value RAM @ DDR400 and he gets 6GB/s on DFI Lanparty (not sure which one) so you're really off on your RAM speed unless you're running Single Channel.

~~Mad Mod Mike, pimpin' the world 1 rig at a time
 
Cheers for the reply. Yeah, you are right about the single channel, I'm getting another module to make it 1GB dual channel as soon as I possibly can. Nice one. Looks like i'll be off to buy some more RAM.
 
I was running a memory test the other day, the one at the beginning of SUSE linux install, and it stated that the L1 Cache was - 16,476MBps, and L2 Cache was - 3,925MBps.

Is this normal? What Frequency do L1 and L2 cache run at?

Also it stated my DDR400 is only 1,602MBps, you and I both know it is supposed to be around 3200,MBps. Maybe the memory test has measured these speeds wrongly? Because I would have thought that all these speeds are suposed to be faster, especially the L2 Cache.

(AMD ATHLON 64 3200 VENICE, ASUS A8N-SLI DELUXE)
Click to expand...

also depends on what apps you use to benchmark, like Memtest will post my P3's ram speed at like ~300mb/s where as sisoft sandra will post ~1000mb/s
 
How come memory tests get results so far off sometimes? Do latencies really affect bandwidth that much?
I am using Corsair ValueSelect DDR400 (CAS2.5, Even though the CAS is pretty good I think the other timings aren't so good, but I'm new to latencies)
The way I was thinking was that DDR400 Single Channel- 3GBps approx
and Dual Channel 6GBps approx. Find it hard to believe that latencies could effect it that much.
 
Hmmm... you should see a theoretical bandwidth of 3.2GB/s in single channel mode. It's strange that the benchmarking program reported 1.6GB/s (exactly half of what it's supposed to be).

Do you have any other memory benchmarking programs? I question the accuracy of the one you're using...

Also, are you 100% positive the RAM is running at the correct clock speed?
 
I am using Corsair ValueSelect DDR400 (CAS2.5, Even though the CAS is pretty good I think the other timings aren't so good, but I'm new to latencies)
Click to expand...

I run 2 GB (4 x 512mb) of Corsair ValueSelect DDR400. Latencies are shown below from CPU-Z:
corsair.PNG


Not too bad for the price of the ram if you ask me.
 
Yeah those latencies pretty much the same as what is says in my BIOS, and the clock is definitely at the right speed.
So yeah it must be a memtest problem
Cheers
 
Hmmm... you should see a theoretical bandwidth of 3.2GB/s in single channel mode. It's strange that the benchmarking program reported 1.6GB/s (exactly half of what it's supposed to be).

Do you have any other memory benchmarking programs? I question the accuracy of the one you're using...

Also, are you 100% positive the RAM is running at the correct clock speed?
Click to expand...

My old celeron 2400 (northwood) with 512mb DDR266 dual channel ram hits 1.5gb/s max in sisoft sandra.

all depends on software used to benchmark.
 
Yeah valueRAM (from almost anyone) is a good deal if you don't plan to overclock.
Click to expand...

I have some value RAM of Kingston that gets higher efficiency than my PC4000 OCZ. 94% vs. 88%.

~~Mad Mod Mike, pimpin' the world 1 rig at a time
 
I would imagine the PC4000 stuff has higher latencies?
Click to expand...

3-3-3-8 (Kingston) vs. 3-3-3-7 (OCZ), the PC4000 is @ 225MHz and the Kingston is @ 184MHz. Same thing with my PC3200 Crucial Ballistix, it got MAX 90% efficiency, but I have the ASUS A8N-SLI and the Kingston is in a DFI Lanparty, I blame the mobo.

~~Mad Mod Mike, pimpin' the world 1 rig at a time
 
I keep meaning to ask... what do video card do you use for your professional work? Quadro I'm guessing- but which one? SLI?
Click to expand...

I wish SLI, lol. The Engineering Lab has the FX5xx Series I believe(The crappy ones) but never cared to look for specifics. I only dropped by there a few times, my "professional" work involves MCSA, but I wouldn't call that "professional" 😵

PS: What kind of efficiency are you getting on your system for RAM? I would think 94% on that Value w/ DFI means ASUS = Sucks.

~~Mad Mod Mike, pimpin' the world 1 rig at a time
 
I use a QuadroFX 3000 in my office. It's starting to show its age with latest version of AutoCAD. My lappy has a QuadroFX Go 700- it's really only good for looking at small wireframe projects.

A QuadroFX 4500 in SLI would be sweeeeeeeeeeeet.

As for my RAM efficiency, I have no clue. I'm on secured computer which means I can't install programs or run very many. I don't even know what brand memory I have on this thing. LOL. It's got so damn many protections. If I open the case it gets reported to the security geeks.

Ironically it's a lenovo! hehehee...

-mpjesse
 
I use a QuadroFX 3000 in my office. It's starting to show its age with latest version of AutoCAD. My lappy has a QuadroFX Go 700- it's really only good for looking at small wireframe projects.

A QuadroFX 4500 in SLI would be sweeeeeeeeeeeet.

As for my RAM efficiency, I have no clue. I'm on secured computer which means I can't install programs or run very many. I don't even know what brand memory I have on this thing. LOL. It's got so damn many protections. If I open the case it gets reported to the security geeks.

Ironically it's a lenovo! hehehee...

-mpjesse
Click to expand...

What about your home PC? I remember you talking about your new comp w/ the Opty 165.

Haha, security is fun. I remember setting up security on some computers that wouldn't let them shutdown, log into a website, or even open paint, haha. I love security ;D.

~~Mad Mod Mike, pimpin' the world 1 rig at a time
 
My home PC? I've run memory benches in the past, but I never kept a score. I think you might be confusing the Opty 165 statement w/ someone else. I've got a X2 4200+

I think I did contemplate getting a Opty 165 for OC'ing... but I never did get it.

Gimme a link to a benchmarking utility and i'll bench my memory when i get home. What u use? Sandra?
 
My home PC? I've run memory benches in the past, but I never kept a score. I think you might be confusing the Opty 165 statement w/ someone else. I've got a X2 4200+

I think I did contemplate getting a Opty 165 for OC'ing... but I never did get it.

Gimme a link to a benchmarking utility and i'll bench my memory when i get home. What u use? Sandra?
Click to expand...

Ah, yes, the 4200+.

Here's a link: http://download.guru3d.com/sandra/

Yea, I use Sandra, it tells the Efficiency but it has a bug with it in 2T where it chops off over 1GB/s bandwidth on your score.

~~Mad Mod Mike, pimpin' the world 1 rig at a time
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom