An overclocking delight, 2300MHz for a good price.
For gaming - except with UT3 - the most rewarding characteristic of memory is probably the overall latency that can be achieved when pairing both high-speeds and low latency. This memory kit from OCZ has it all:
- OCZ3P1600LV6GK
- 240Pin Unbuffered-DIMM Non ECC
- 6GB Triple Channel Kit (2GB x3)
- DDR3 1600
- 7-7-7-24
- 1.65v
Lab501 tested the said memory and was able to achieve Super Pi benchmarking at DDR3 2300 9-9-9-24 1T @1.90v speeds, while suicide shots could be had at 2400MHz, with the same settings.
OCZ Platinum DDR3 1600 @ Lab501
4 comments:
Something seems a bit off on this one.
Firstly, i thought i7 boards maxed-out on 1.65v for the RAM.
Secondly, how have you possibly got an AGP graphics card in a Core i7 MB?!!!
Hi Philip,
Actually they are using a PCI GF440MX, which I didn't know it existed but I have still saw some new cards for PCI bus a year ago or so, GeForce 8400, 8500 and 8600.
As for the RAM voltage, Intel doesn't sanction more than 1.65v but, with the right tunning, it's possible to achieve that without burning the CPU's memory controller. The main setting is to keep Uncore/QPI voltage within 0.5v of the DRAM voltage, never with a bigger difference than that, and you should be fine. Of course that for 24/7, you wouldn't want anything so overclocked either way.
The max intel recommends for QPI is 1.35v, so that takes you up to 1.85v DRAM voltage while not breaking the 0.5v difference.
hi Tiago
i am sure you are right but it does say AGP8x on the screen grab.
thanks for the advice on the voltages. when you say about not running it over-volted 24/7 is that purely because of the heat? ie if you ran your conputer in the Arctic and all the temps were well under what they normally are at room temp. stock voltages then you would/should have no problems running a well overclock system for several years?
Phil
Hi Philip,
I know it does, must be an error of some sort, due to the unexpected configuration. Nobody uses those PCI cards anymore, except for overclocking like crazy. I was also surprised to see that ;)
Heat is not normally the problem, as the thermal fatigue would only cause you problems years down the road. The usual problem is electromigration, which causes the silicon paths inside the chip to start to desintegrate from it's normal structure, which ends in the material loosing it's fabricated properties - the transistors stop behaving as such.
See:
http://en.wikipedia.org/wiki/Electromigration
This is exacerbated by high currents and high voltages and also increases as the process node shrinks.
If this is the cause, it would be something to expect since the i7 is manufactured at 45nm, considerably smaller than Intel's Core 2 chipsets, increasing this effect. The DDR3 ram is also more current demanding, further increasing the problem.
Electromigration was the cause of the sudden death of some P4 processors some years ago, while overclocked.
In the case of the i7, it might not be electromigration related since people have not been having problems when keeping VTT/QPI voltage within the 0.5v difference.
What actually is the problem, only Intel knows.
For reference, voltages as high as:
1.6v vCore, 1.6v VTT/QPI and 2v VDIMM
have not killed CPUs yet.
If you can have the luxury of replacing your i7 in case something goes wrong, keep VDIMM below 1.65v, as it's the only certified voltage from Intel.
As for heat, while it won't kill your CPU in a short timespan, try to react to it's increase accordingly when you push voltages up. A cooler processor is one that overclocks more.
Best regards,
Tiago
Post a Comment