DDR2 Bandwidth Benchmarks: Post Here

SiSoftware Sandra

Benchmark Results
RAM Bandwidth Int Buff'd iSSE2 : 7112 MB/s
RAM Bandwidth Float Buff'd iSSE2 : 7014 MB/s
Results Interpretation : Higher index values are better.

Int Buff'd iSSE2 (Integer STREAM) Results Breakdown
Assignment : 7211MB/s
Scaling : 7175MB/s
Addition : 7044MB/s
Triad : 7020MB/s
Data Item Size : 16 byte(s)
Buffering Used : Yes
Offset Displacement Used : Yes
Bandwidth Efficiency : 81% (estimated)

Float Buff'd iSSE2 (Float STREAM) Results Breakdown
Assignment : 7162MB/s
Scaling : 7174MB/s
Addition : 6898MB/s
Triad : 6823MB/s
Data Item Size : 16 byte(s)
Buffering Used : Yes
Offset Displacement Used : Yes
Bandwidth Efficiency : 80% (estimated)

Performance Test Status
Run ID : NICHOLAS-94BE82 on Friday, October 21, 2005 at 8:27:49 PM
Memory Used by Test : 512MB
NUMA Support : No
SMP Test : No
Total Test Threads : 1
Multi-Core Test : No
SMT Test : No
Dynamic MP/MT Load Balance : No
Processor Affinity : P0C0T0
Page Size : 4kB
Use Large Memory Pages : No

Chipset 1
Model : Advanced Micro Devices (AMD) Athlon 64 / Opteron HyperTransport Technology Configuration
Front Side Bus Speed : 2x 1100MHz (2200MHz data rate)
In/Out Width : 16-bit / 16-bit
Maximum Bus Bandwidth : 8800MB/s (estimated)

Logical/Chipset 1 Memory Banks
Bank 2 : 512MB DDR-SDRAM 2.5-4-4-8 2CMD
Bank 3 : 512MB DDR-SDRAM 2.5-4-4-8 2CMD
Channels : 1
Bank Interleave : 2-way
Speed : 2x 275MHz (550MHz data rate)
Width : 128-bit
Maximum Memory Bus Bandwidth : 8800MB/s (estimated)

Features
(W)MMX Technology : Yes
SSE Technology : Yes
SSE2 Technology : Yes
SSE3 Technology : No
EMMX - Extended MMX Technology : Yes
HTT - Hyper-Threading Technology : No

Performance Tips
Notice 5008 : To change benchmarks, click Options.
Notice 5004 : Synthetic benchmark. May not tally with 'real-life' performance.
Notice 5006 : Only compare the results with ones obtained using the same version!
Tip 2 : Double-click tip or press Enter while a tip is selected for more information about the tip.


I've never bothered testing the memory above 275x2 2.5-4-4-8 although I'm sure it's capable. It's G.Skill PC4400, and DDR1, for shame.

btw it's on crack about 2t command rate and single channel.... everything is set up correctly in BIOS.. dunno why it says single channel ..

SuperFarStucker said:

btw it's on crack about 2t command rate and single channel.... everything is set up correctly in BIOS.. dunno why it says single channel ..

Click to expand...

Thanks for the info but I was hoping to keep the thread to DDR2 benchmarks.

batboy said:

I just happen to have 2X512 Samsung single sided PC2-3200 RAM that is going in my Mom's rig. If I have time tonight, I'll try that for you. I'll also try a couple max runs with my good DDR2 RAM too. Someone needs to give Crimedog some competition.

Click to expand...

Looking forward to the results and differences!

That's great bandwidth out of DDR2. But just out of curiosity, what kind of timing you’re getting out of Super pi 32M, or Everest's Latency? Would be a good Idea to show those benches too, and see how it's comparing to DDR.
A lot of Bandwidth, but how fast can the DATA get to CPU and Back? That is the question. I mean do we have a 12 lanes highway for a Farm tractor or can I drive my Porsche 300 mph?

I am curious why my plain ddr at 306 FSB apparently beats all but 2 of you in the bandwidth ? I thought DDR2 was better ram ..

http://img.photobucket.com/albums/v87/diehrd/236778a3.jpg

And at 315 FSB I beat all but 1 of ya ..

http://img.photobucket.com/albums/v87/diehrd/Image315fsb.jpg

So what is the point to ddr2 if any one knows ?

Even w/ my ram @ 400mhz, 4-4-4-12 I only get around 6400mb in Sandra, would you say the 2:3 divider is killing me?

diehrd said:

So what is the point to ddr2 if any one knows ?

Click to expand...

Lower voltages for starters. As the chipsets improve so has the efficiency/bandwidth. Took a while for DDR to get to where it's at.

Hmmm ok thats cool a lower voltage Ram i am at 2.7V on mine..But aside from that I guess there is little or no benifit ?

Poor me, I only could get 2700mhz out of my A64 3000 Venice. My SISandra is a mere 6899, and Everest Latency is a Giant 39.2 ns. I know.. This is DDR2 Thread and for the big boys. But please show me your Everest latency and make me more miserable.
http://img434.imageshack.us/my.php?image=102220051039497ev.jpg

Using 5/6 divider increased bandwidth and decreased Latency (to 38.6 ns)

http://img121.imageshack.us/my.php?image=102220051604553gs.jpg

wow my everest latency is terrible, i don't think this can be right. anyone have some info about ddr2 and everest?

edit: i think it's an everest thing. i ran read and it was 9100mb/s, write was around 2800mb/s, and that was at 333mhz 1:1

No, Everest isn't screwing anything up, I don't think. It's common sense that a Athlon64 with an integrated memory controller and (probably) lower latency RAM would beat you out by a good margin in anything latency-related. You're going through the chipset still, and those A64 rigs aren't.

In my mind, this game is a lot like hard drives. Show off all the sequential read speeds (bandwidth) you want, but a 7200RPM drive isn't going to come close to a Raptor in terms of seek times (latency).

johan851 said:

In my mind, this game is a lot like hard drives. Show off all the sequential read speeds (bandwidth) you want, but a 7200RPM drive isn't going to come close to a Raptor in terms of seek times (latency).

Click to expand...

i have used both and i don't think so

johan851 said:

No, Everest isn't screwing anything up, I don't think. It's common sense that a Athlon64 with an integrated memory controller and (probably) lower latency RAM would beat you out by a good margin in anything latency-related. You're going through the chipset still, and those A64 rigs aren't.

Click to expand...

i doubt it makes that big a difference

crimedog said:

i have used both and i don't think so


i doubt it makes that big a difference

Click to expand...

At this juncture it is hard to figure out whether it is so much DDR2 high latency (tCL), or Intel's front side bus, or both causing such high latency. When AMD M2 enters the ring, we will know how good DDR2 is vs DDR, or whether M2 any improvement in memory.
If DDR2 had low timing as DDR, it would be a quite improvement to make the change, but for me I wait untill I know changing would be worth my money.

i doubt it makes that big a difference

Click to expand...

I doubt you read my post correctly. *shrug* Why so short?

What I meant was that a lower latency controller with lower latency RAM will produce lower latencies in a latency benchmark. Can't dispute that.

i have used both and i don't think so

Click to expand...

I'm not comparing performance, I'm commenting on the differences between bandwidth and latency. How they interact in a real world scenario will, of course, be much different than the way hard drives interact. It doesn't change the fact that it's basically the same principle - access time vs. transfer time. K?

Everyone is so defensive about their rigs around here...I don't get it. They're just computers. Or maybe it's too late.

*Edit* Yeah, it's pretty late.

By request, here are the Sandra buffered memory bandwidth results of single sided vs. double sided DDR2.

I clocked my system way down so I could run the lower speed RAM (that's why the score is so low). System is running at 4 gig (211 FSB). RAM frequency is 210.7 using the 1:1 ratio (DDR2-421.4) and timing is a relaxed 4-4-4-12.

Samsung PC2-3200 single sided 2X512... Int=4527 and Float=4525

Kingston HyperX PC2-5400 single sided 2X512... Int.=4545 and Float=4515

OCZ EB Patinum Edition Rev. 2 PC2-4200 double sided 2X512... Int=4514 and Float=4511

Conclusion: looks like they are all about equal. Differences are easily within the margin of error.

johan851 said:

Everyone is so defensive about their rigs around here...I don't get it. They're just computers. Or maybe it's too late.

Click to expand...

It's not just how late the time is.

People in general are resistent to change. It's also hard to convince someone who just spent $200 on DDR that DDR2 is decent stuff. Add to that the fact DDR2 will be the standard beginning next year (rendering their recent $200 purchase obsolete in future motherboards/upgrade routes). You could say it's partially resentment towards an ever-changing industry. Staying on top could break the bank. It does mine! lol Oh, and we're mostly geeks. We take technology a tad too seriously by nature. PC components are part of the family.

I've been around here for a few days , and have seen a fair number of changes.

johan851 said:

I doubt you read my post correctly. *shrug* Why so short?

What I meant was that a lower latency controller with lower latency RAM will produce lower latencies in a latency benchmark. Can't dispute that.


I'm not comparing performance, I'm commenting on the differences between bandwidth and latency. How they interact in a real world scenario will, of course, be much different than the way hard drives interact. It doesn't change the fact that it's basically the same principle - access time vs. transfer time. K?

Everyone is so defensive about their rigs around here...I don't get it. They're just computers. Or maybe it's too late.

*Edit* Yeah, it's pretty late.

Click to expand...

i'm not saying that ddr2 is faster, i'm not saying it's slower. i just think that benchmark is flawed because i'm running ddr2 really-really fast and it's slaughtered by ddr1 @ stock. if it's not flawed then yes, that's obviously why, but that wasn't my point. don't feel any need to defend my hardware for e-penis status.

Here it goes. Not the greatest, but it is A LOT better than my other rig.

Everest definitely favors AMD systems. Remember when Intel totally dominated memory benchmarks? Do a search and see how many posts you can find that AMD people claimed memory bandwidth benchmarks didn't mean anything. Admittedly, Sandra memory bandwidth benchmarks are sadly out of date.

batboy said:

By request, here are the Sandra buffered memory bandwidth results of single sided vs. double sided DDR2.

I clocked my system way down so I could run the lower speed RAM (that's why the score is so low). System is running at 4 gig (211 FSB). RAM frequency is 210.7 using the 1:1 ratio (DDR2-421.4) and timing is a relaxed 4-4-4-12.

Samsung PC2-3200 single sided 2X512... Int=4527 and Float=4525

Kingston HyperX PC2-5400 single sided 2X512... Int.=4545 and Float=4515

OCZ EB Patinum Edition Rev. 2 PC2-4200 double sided 2X512... Int=4514 and Float=4511

Conclusion: looks like they are all about equal. Differences are easily within the margin of error.

Click to expand...

Thanks for the benches.

Seems like they are all equal on Buffered. What about Unbuffered run? Any idea?