Overclock.net › Member Blogs › Supertalent 2000cl8 My Experience Part 2

Supertalent 2000cl8 My Experience Part 2

Software Used:

AIDA64: Cache and Memory Benchmark


Purpose: To determine and compare the latencies and benchmarks of the RAM at different speeds and settings.
A synthetic test.

CPU-Z:

Purpose: To show the MemOC and the CPU OC

Linx: To determine system stabillity. 10 passes will be used per oc as I intend on doing this many times.

Prime95 Blend:

Purpose:
To determine RAM stabillity and wether it is error free or not. 4+ Hours = A pass in this review.

memtest86+ : 2 Passes on each test will be used in this review, to see if the ram presents errors at any given speed.

MaxxMEM: My personal favorite benchmark. Used to compare ram speeds at different timings.

SuperPI32M: Stabillity test.

Super PI 1M: Speed test / comparison.


In order for a FREQUENCIES/TIMING Combination to be posted in this review, it MUST pass all these tests using 1.65v-1.7v.
This is to give an average user a good idea of what speeds he can get without overvolting their DIMM.

Overclocking:

Overclocking hypers is something that has been well documented. If you are looking to see how they scale with voltage, there are plenty of places to find that. This Overclocking test was to determine how much extra juice you can get out of your STT's without going over 1.65V.

Again, this isn't extreme overclocking by any stretch. All these clocks were attained using 1.65V.

Please note, all of my stable clocks thus far have been achieved using the 2x Uncore rule. You may find that the amount of Uncore you must give your IMC to stabilize your RAM OC to require a little more then x2 the ram frequency. I have not found this to be the case in any situation, but each chip has different IMC requirements.

So, using 1.65V, we were able to attain the following stable overclocks on this ram:

1600 6-6-5-16-1T TRFC 48 1.65V *STABLE

1600 6-6-6-18-1T TRFC 48/72 1.65V * STABLE

1700 7-7-6-24-1T TRFC AUTO/72 1.65V *
STABLE

1754 7-7-6-24-1T TRFC AUTO/72 1.65V * STABLE

2000 8-8-8-24-1T 1.65V *STABLE

2000 8-8-8-24-2T 1.65V *STABLE

2000 8-7-8-24-1T 1.65V *STABLE


These are some very impressive numbers indeed! the tRCD speed at 2000 could stably be set to 7 without any extra VDIMM!!!!
These are the clocks that I have personally tested to be stable on this ram, just so you have a good idea of what you can do for a 24/7 setup.

The Benchmarks will cover a multitude of speed/timing combination(s) in order to give a good comparison between the different speeds, but these are the most common 24/7 clocks you are likely to get/use with this ram.

Overclocking Conclusion:

These sticks did a great job. You can get so much more juice from them with a mere 1.65V. I was able to lower the tRCD timing to 7 in 2000MhZ stably, with no extra voltage required. Not to mention, the ram had no problem running in 1T timing mode, as opposed to the XMP defaulted 2N. Very nice results for 2000.

1600 is where I personally think these shine. I was able to get a stable clock of 1600 6-6-5-16-1T with a tRFC of 48. This is stable in Prime95, Linx(10 passes), Memtest86+, and SuperPI32M stable.

Have a look at the difference between 1600 6-6-5-16-1T and 2000CL8.

Although the 2000CL8 has a lower latency and slightly higher bandwidth, the difference is certainly not large enough to justify an increase in QPI/VTT voltage.

So if you plan on Overclocking these bad boys, then you will sure have a blast. Just imagine how they would scale with more voltage.
Lets take a look at how they performed on benchmarks.

Benchmarking:

If you have read this far into the review, you are probably interested in how well these sticks can overclock with no added voltage. Although there was no drastic increases made to the RAM frequencies in the overclocking section, it was more of a demonstration on how tight you could go with the timings and how well the timings scaled when the frequency was lowered.

Lets take a look at how the different ram speeds performed at different benchmarks in comparison with each other:

Latency Benchmark:
Software: AIDA64 Cache and Memory Benchmark, Latency (NS)

Latency Benchmark 2: MaxxMEM:

Bandwith Benchmark:
Software: AIDA64 Cache and Memory Benchmark, READ/WRITE/COPY.
Bandwith Benchmark 2: MaxxMEM


Benchmark 3
:
MaxxMEM Scores
Benchmark 4:
superpi1m,32m runs

Conclusion:
Well, hopefully not only the stable speeds they were capable of doing on 1.65VDIMM helped make a decision for you. The benchmarks were very impressive all around, doing some very low latencies.

The speed benefit is clear. These kits are simply able to do speeds / timing combination's no other kits presently available for purchase are able to do. The benchmarks are favorable, all of which we managed to achieve very nice scores in.

So, if you are looking for some damn good RAM that scales, overclocks, under clocks, and just in general performs really really well, then look no further. If you want what is ostensibly some of the fastest memory available for purchase then look no further.

This memory is for those who want:

- Great Overclockability

- Great synthetic scores

- Some of the tightest speeds you can find at 2000 / 1600 MHz using stock voltage

- Great Scalability with more voltage

Comments

There are no comments yet
Overclock.net › Member Blogs › Supertalent 2000cl8 My Experience Part 2