Ram speed higher , bench results lower ?

Gotta be honest while I love this kit - its probably not he best suited to people who are doing stuff thats performance bias towards higher RAM bandwidth.

I got them purposefully to tighten up the timings as most of the stuff I do I know gets better performance from a bias towards tighter timings while still maintaining good bandwidth which they do well at 2133.
 
Below are shots of the auto secondary and thirds on the vulcan reds for comparison.




Ive had them upto 2600mhz before, timings at 11-13-12-31 1T, 1T. TRFC at 196 and tRDRD at 6 along with the last three 3rds at 12. Info passed on by 8Pack in a thread i made some time ago.
 
Try flatten out tRP and drop CL, mine didn't like dropping back off on tRP. Try these at 2400mhz (remember to nudge up SA/IOA/IOD/PCH/DRV too!)

10-12-12-32-114-1T (slack)
9-12-12-32-96-2T (tight)
9-11-11-28-96-1T (tighter)

Don't be scared to fire the dram volts up towards 1.7v either.
 
Thanks guys , much appreciated and insightful stuff. @setter - your xmp seems much lower clocks from what the beast provides. For instance your TRFC shows 193 clocks ,, on these its 312 at XMP

@PGI - nice pointers. though at the moment i have zero movement on those main timings. It just does not boot. Im thinking possibly its lack of voltage bumps in the correct places as you mention. could you explain these "(remember to nudge up SA/IOA/IOD/PCH/DRV too!) " in full form and ill apply go again tomorrow.

Thanks to all. ;D
 
This is the second 2400mhz kit ive had, the first was a patriot venom red c11 kit, though it turned out to be faulty. Got the team stuff as a replacement, same price as the patriot but better primary timings. Seems that ocuk only stock the c11 kits now.
 
setter said:
This is the second 2400mhz kit ive had, the first was a patriot venom red c11 kit, though it turned out to be faulty. Got the team stuff as a replacement, same price as the patriot but better primary timings. Seems that ocuk only stock the c11 kits now.
Click to expand...

Does seem to be the case setter :( Though TG clearly give something a little extra in there kits. ;)
 
McstylisT said:
Thanks guys , much appreciated and insightful stuff. @setter - your xmp seems much lower clocks from what the beast provides. For instance your TRFC shows 193 clocks ,, on these its 312 at XMP

@PGI - nice pointers. though at the moment i have zero movement on those main timings. It just does not boot. Im thinking possibly its lack of voltage bumps in the correct places as you mention. could you explain these "(remember to nudge up SA/IOA/IOD/PCH/DRV too!) " in full form and ill apply go again tomorrow.

Thanks to all. ;D
Click to expand...

Max voltage range for these are usually 1.2/1.22v, though you should only really need at most 1.15-1.18v. Also noticed you're on an ITX board, you may have less oomph to give the memory.

SA = System Agent, sometimes known as VCCSA
IOD = I/O Digital
IOA = I/O Analogue
PCH = Platform Controller Hub

DRV = Dram Voltage, should default to 1.65v under XMP conditions, no harm in bumping this to 1.68-1.7v

http://www.overclock.net/t/1401976/the-gigabyte-z87-haswell-overclocking-oc-guide - A good guide with some charts in as to which voltage settings help overclocking which ever part of your system :)
 
Running 1.2 on SA, IOD/A and PCH here. Anything lower results in 1A memory management bsod's.
 
pgi947 said:
That can't be helping your temps bud, PCH is killer for high CPU temps.
Click to expand...
Ill see about lowering PCh a tad, however though ive recently had to up cpu voltage due to whea errors and related 0x124 bsod's in games. But even before that this has been an extremely hot chip. 1.265v under p95 fft resulted in 94c on one core inside 3 seconds.
 
ouch, thought about dropping the clock a little? 4.3-4.4 is still decent (running around 5ghz in sandy terms). What do you have uncore and ring voltages set at?
 
On asus theyre listed differently mate. Uncore is known as cache, which ive left on auto as im running at the stock 39x. Cpu input voltage is the other, currently at 1.85, according to guides this is ok upto 1.200. Temps are only high if i have the heating on, gaming today its maxing at 67c under a pretty aggressive cpu fan profile.
 
You mean 1.185? :p

Try dropping the cpu multiplier and raise cache multiplier, perfecting efficiency on haswell can be far more beneficial then going for raw mhz :)
 
pgi947 said:
You mean 1.185? :p

Try dropping the cpu multiplier and raise cache multiplier, perfecting efficiency on haswell can be far more beneficial then going for raw mhz :)
Click to expand...
Nope, 1.85. Basically it's the voltage applied at post, VRIN on a gigabyte. It's explained better here.

http://rog.asus.com/254052013/maxim...ngs-for-overclocking-maximus-vi-motherboards/

The terms for settings between different manufacturers can be a bit confusing. Ill see about dropping the multi a tad, though i have ran before with 45x cpu multi and cache at 43x on 1.200v. Certainly helps in a few benchmarks.
 
Yep, imho there should be an industry standard for theese terms. VRIN and VRING in particular cause an awful amount of confusion.
 
Yes, especially as if your set VRING to VRIN voltages, you're going to have a bad time :p

It was always the whole VCCSA/SA/IMCV/VTT naming that threw me.
 
You must log in or register to reply here.
Back
Top Bottom