|
Testing:::...
Test System
- Abit IC7-Max3 Mainboard
- 2.8C Pentium 4
- 2 Seagate 7200.7 80GB 8MB Cache drives configured for RAID 0
- EVGA 6800GT with Artic Cooling VGA Silencer
First off, a few notes. After doing some research, I’ve discovered that this ram is voltage hungry. As mentioned earlier, these modules sport BH-5 chips, which is the reason for the hunger for the voltage. I only have one system that can do higher voltage levels and this is the board listed above, an IC7-Max-3. The problem is that the IC7-Max 3, while still a great board for its day, has well known issues with voltages over 2.9, because of a vtt tracking issue, and requires a mod to rectify this issue. One I’m not going to attempt, as its a little over my head. It’s also well known that the newer systems on the market do have much higher voltages to play with. Having said that, I was still able to run off a few tests, and even went up to 3.3V, and came off with some interesting results.
We are going to be comparing the OCZ kit to a PDP Systems Patriot 1GB PC3200 Dual Channel Kit. It also is billed as a low latency ram. I merely threw it in there for comparison proposes. There are 3 sets of results to look at with these numbers. 2 of them are both sets of Dual Channel Kits at the same voltage, speeds, and timings. The third set is the OCZ as far as I could push it on my system, which was 2-2-2-5 @ 230. I do believe that this ram can go higher, and I was able to get it to boot up at 240, but the cost was some instability. However. Since this board does have issues supplying proper voltages in the higher range, and since I do not know the full limit of my particular processor, it would be incorrect of me to even try to come to a good conclusion as to its upper FSB limits with what I have here. I did however manage to get some solid numbers off of a good over clock session.
I had hoped to include the Patriot ram in the overclocked results; however, I found that that ram wouldn’t do the same speeds as the OCZ when pushed. Even with the voltage up, and timings set to 3-4-4-8! So I opted to exclude the data. At the least, you can use similar speeds, timings and voltages for the first set of numbers for comparison, while the third gives you an idea of where the OCZ ram can go when pushed. Ok, enough of my ramblings, on to the numbers!
ScienceMark 2:-
From the ScienceMark site - Science Mark 2.0 is an attempt to put the truth behind benchmarking. In an attempt to model real world demands and performance, SM2 is a suite of high-performance benchmarks that realistically stress system performance without architectural bias.
->Patriot 2.5-3-2-5 | and 2.7V 200FSB
Memory Bandwidth 4053.02 MB/s
Latency
- 4 Cycles / 1.43 ns / 4 byte stride
- 15 Cycles / 5.34 ns / 16 byte stride
- 55 Cycles / 19.60 ns / 64 byte stride
- 246 Cycles / 87.66 ns / 256 byte stride
- 225 Cycles / 80.17 ns / 512 byte stride
->OCZ 2.5-3-2-5 | and 2.8V 200FSB
Memory Bandwidth 4090.33 MB/s
Latency
- 4 Cycles / 1.43 ns / 4 byte stride
- 15 Cycles / 5.34 ns / 16 byte stride
- 54 Cycles / 19.24 ns / 64 byte stride
- 240 Cycles / 80.52 ns / 256 byte stride
- 231 Cycles / 82.31 ns / 512 byte stride
->OCZ 2-2-2-5 | and 2.8V 200FSB
Memory Bandwidth 4189.91 MB/s
Latency
- 4 Cycles / 1.43 ns / 4 byte stride
- 15 Cycles / 5.34 ns / 16 byte stride
- 55 Cycles / 19.60 ns / 64 byte stride
- 222 Cycles / 79.10 ns / 256 byte stride
- 234 Cycles / 83.38 ns / 512 byte stride
->OCZ_OC'D 2-2-2-5 | and 3.2V 230FSB
Memory Bandwidth 4703.82 MB/s
Latency
- 4 Cycles / 1.24 ns / 4 byte stride
- 15 Cycles / 4.65 ns / 16 byte stride
- 54 Cycles / 17.06 ns / 64 byte stride
- 240 Cycles / 71.65 ns / 256 byte stride
- 231 Cycles / 63.90 ns / 512 byte stride
The ScienceMark2 results show a slight increase in performance by the OCZ over the Patriot when the settings are set the same. As you will see, this is a recurring theme. The OCZ took a nice leap when pushed. Notice the latency drop in the 512 byte stride results for the pushed OCZ numbers, which wasn’t present with the lower settings. Leading me to feel as though this ram is hungry for higher voltages and speeds.
EVEREST:-
From Lavalys' site - EVEREST Home Edition is a freeware1 hardware diagnostics and memory benchmarking solution for home PC users, based on the award-winning EVEREST Technology. It offers the world's most accurate hardware information and diagnostics capabilities, including online features, memory benchmarks, hardware monitoring, and low-level hardware information.
->Patriot 2.5-3-2-5 | and 2.7V
- Read 4751 MBs
- Wite 1541 MBs
- Latency 81.6ns
->OCZ 2.5-3-2-5 | and 2.8V
- Read 4824 MBs
- Wite 1543 MBs
- Latency 81.0.8ns
->OCZ 2-2-2-5 | and 2.8V
- Read 4859 MB\s
- Write 1565 MB\s
- Latency 79.8 ns
->OCZ_OC'D 2-2-2-5 | and 3.2V 230FSB
- Read 5568 MBs
- Wite 1754 MBs
- Latency 76.1.8ns
As you can see from the Everest results, yet again the OCZ edges out the Patriot when the timings and speeds are equal. The OCZ made a big jump when pushed.
SiSoft Sandra:-
->Patriot 2.5-3-2-5 | and 2.7V 200FSB
- Ram Bandwidth Int Buff'd iSSE2 4783 MBs
- Ram Bandwidth Float Buff'd iSSE2 4818 MBs
->OCZ 2.5-3-2-5 | and 2.8V 200FSB
- Ram Bandwidth Int Buff'd iSSE2 4844 MBs
- Ram Bandwidth Float Buff'd iSSE2 4865 MBs
->OCZ 2-2-2-5 | and 2.8V 200FSB
- Ram Bandwidth Int Buff'd iSSE2 4888 MB\s
- Ram Bandwidth Float Buff'd iSSE2 4865 MBs 4892 MB\s
->OCZ_OC'D 2-2-2-5 | and 3.2V 230FSB
- Ram Bandwidth Int Buff'd iSSE2 5639 MBs
- Ram Bandwidth Float Buff'd iSSE2 5636 MBs
Sandra results give the same impression as the others, the OCZ moving ahead slightly when things are equal, and a big jump when pushed.
FM PCMark05 Memory Suite Bench:-
->Patriot 2.5-3-2-5 | and 2.7V 200FSB
4132
->OCZ 2.5-3-2-5 | and 2.8V 200FSB
4151
->OCZ 2-2-2-5 | and 2.8V 200FSB
4116
->OCZ_OC'D 2-2-2-5 | and 3.2V 230FSB
4784
FutureMarks results with equal settings are perhaps more close, but yet again we see a leap when pushed.
|
