|
Page 2 of 2
Benchmarks:
We ran the test rig at 17 x 800Hz FSB and at 17 x 932MHz FSB and these are the results using Sandra Professional 2004. The memory really pulls through when over clocking as you can see in the configuration 2 test. While the system was tested at default, we found the bandwidth efficiency signiticant over other DDR2 Solutions with others averages 74%. When we pumped up the bus the ram took off and kicked out some impressive scores over other 4-4-4-12 ram we have tested.
Configuration 1 Model : Intel Corporation 925x Memory Controller Hub Front Side Bus Speed : 4x 200MHz (800MHz data rate) Width : 64-bit Maximum Bus Bandwidth : 6400MB/s (estimated)
Logical/Chipset 1 Memory Banks Bank 0 : 512MB DDR2-SDRAM 3.0-3-3-8CL 1CMD Bank 4 : 512MB DDR2-SDRAM 3.0-3-3-8CL 1CMD Channels : 2 Width : 64-bit Performance Acceleration Technology : Yes

SiSoftware Sandra

|
Sandra Benchmark |
Results |
|
Int Buff'diSSE2 |
4924 MB/s |
|
Float Buff'd iSSE2 |
4929 MB/s |
| Maximum Bus Bandwidth |
7456MB/s |
Int Buff'd iSSE2 (Integer STREAM) Results Breakdown
|
Sandra Benchmark |
Breakdown |
|
Assignment |
4792MB/s |
|
Scaling |
4792MB/s |
| Addition |
5057MB/s |
| Triad |
5055MB/s |
| Data Item Size |
16-bytes |
| Buffering Used |
Yes |
| Offset Displacement Used |
Yes |
| Bandwidth Efficiency |
77% (estimated) |
Float Buff'd iSSE2 (Float STREAM) Results Breakdown
|
Sandra Benchmark |
Breakdown |
|
Assignment |
4800MB/s |
|
Scaling |
4792MB/s |
| Addition |
5066MB/s |
| Triad |
55059MB/s |
| Data Item Size |
16-bytes |
| Buffering Used |
Yes |
| Offset Displacement Used |
Yes |
| Bandwidth Efficiency |
77% (estimated) |
Configuration 2 Model : Intel Corporation 925x Memory Controller Hub Front Side Bus Speed : 4x 233MHz (932MHz data rate) Width : 64-bit Maximum Bus Bandwidth : 7456MB/s (estimated)
Logical/Chipset 1 Memory Banks Bank 0 : 512MB DDR2-SDRAM 4.0-4-4-12CL 1CMD Bank 4 : 512MB DDR2-SDRAM 4.0-4-4-12CL 1CMD Channels : 2 Width : 64-bit Performance Acceleration Technology : Yes

SiSoftware Sandra

|
Sandra Benchmark |
Results |
|
Int Buff'diSSE2 |
5775 MB/s |
|
Float Buff'd iSSE2 |
5770 MB/s |
| Maximum Bus Bandwidth |
7456MB/s |
Int Buff'd iSSE2 (Integer STREAM) Results Breakdown
|
Sandra Benchmark |
Breakdown |
|
Assignment |
5623MB/s |
|
Scaling |
5617MB/s |
| Addition |
5931MB/s |
| Triad |
5931MB/s |
| Data Item Size |
16-bytes |
| Buffering Used |
Yes |
| Offset Displacement Used |
Yes |
| Bandwidth Efficiency |
77% (estimated) |
Float Buff'd iSSE2 (Float STREAM) Results Breakdown
|
Sandra Benchmark |
Breakdown |
|
Assignment |
5619MB/s |
|
Scaling |
5616MB/s |
| Addition |
5925MB/s |
| Triad |
5920MB/s |
| Data Item Size |
16-bytes |
| Buffering Used |
Yes |
| Offset Displacement Used |
Yes |
| Bandwidth Efficiency |
77% (estimated) |
Here we can see that DDR leaves off where DDR2 takes over on intel based systems. With a significant 77% efficiency at both configurations, shows this ram crunches data extremely well.

RightMark Memory Analyzer:
RightMark Memory Analyzer reflects the CAS# timings of 3.0, 4.0 & 5.0. While this ram runs at 3.0-3-3-8 you must manually set the timings. We set ours to 17 x 200MHz and snapped this report off. Our real time results put this ram under RightMark Memory Analyzer v3.41 scrutiny and what we came up with is a maximum ram bandwidth of 2959.54 MB/s with a sustained throughput of 2622.33 MB/s of performance bandwidth.




CPU-Z:


While at 200MHz FSB the ram would in deed run at advertised rates however the ram would force a ratio of 3:4 and we could not get a 1:1 ratio past 17 x 233Mhz. While further testing on other platforms and configurations might yield something different, we just have to go with our current test bed reports.

Final Thoughts:
PQI offers super fast ram with elegance and style, the spreaders dont make the ram, but the ram makes the modules and with this kind of performance and looks, PQI will be hitting the retail market like gangbusters. The modules run extremely well and while they look pleasing to the eye, the modules should mature in the future with fine tuned circuits and programmed SPD logic to allow more performance and functionality to future revisions.
Having been said the ram is still fast and stable but we prefered to see the modules default to 3-3-3-8 on a 200MHz FSB configuration. Having to manually set the timings is no big deal as enthusiasts are constantly tweaking there system, but on the other hand, this might be a pain in the butt for others who prefer a pre-programed SPD feature, so we had to give it a negative for that aspect..
The modules for the most part ran at 4-4-4-12 almost the entire analysis with the exception of our first 200MHz FSB benchmark. Our system ran at 4-4-4-12 for almost every increase in FSB we added to the system, we had to losen the timings in order to post when we forced a 233 x 17.0 configuration to a 4-4-4-15. However the ram is so efficient the maximum bandwidth was significantly higher on this system than other 4-4-4-12 offerings.
Intel systems handle loose timings a lot better than AMD so for those who are shopping for good intel ram, keep in mind that 4-4-4-12 is fast for 1st generation ram. We did manage to get a nice 3-3-3-8 timing at 200 x 17.0 and pulled off a stable benchmark and proceeded to use the system with no hitches what so ever. So the system is definately stable at this configuration on our rig.
While we can set manual timings down to a very tight timing of 2-2-2-5 at DDR400 rate, we find that having to adjust the ram is a hit and miss ordeal in that if you go to tight the ram will lock up and it takes a few chances at finding that rams breaking point at any given FSB. Thus we feel the eeprom should be pre-programmed to operate at other frequencies and timings that would provide optimized settings at default.


Pro's:
- SPD must be manually set to 3-3-3-8 Rating.

PQI Corporation 46716 Lakeview Blvd. Fremon, CA 94538 Tel: 510-651-7281 Fax: 510-651-7240
<< Start < Prev 1 2 Next > End >> |