Gigabyte Z97SOC Force legacy IC tests

[websmile]

Hi and welcome to my small thread

 

Lately I remembered how I started overclocking, and that it was on a socket 754 Gigabyte board with Nforce3 chipset. Seeing that i haven´t used a GA board for nearly four years now, I decided that it may be a good decidion to end the oc era the way I started it, and I bought a Gigabyte Z97X SOC Force retail board to do some tests.

This is what I got, the imc of this random Pentium 3258K is quite weak, but hopefully I can give you a small overview of what the board can do on mem oc and that I indeed had fun

 

Hardware used:

Gigabyte Z97X SOC Force retail, Bios x04 and f6 loaded from internet

Intel Pentium G3258K retail

Megalhalems cpu cooler

Crucial SSD

2x2GB G.Skill 2400c9 retail(PSC), 2x2GB G.Slill RipjawX 2133C8 retail(BBSE) and 2x2GB Corsair Dominator GT 2000 7-8-7-20 retail(Hyper), G.Skill RipjawZ 2x4GB 2400c9(Samsung) for imc and profile tests only

Mushkin PSU

 

First I started with f6 Bios which the board was delivered with, unfortunately I wasn´t able to boot twcl6 with any mems and ics, because of this made a fast test on cpu abilities and then changed to Bios X04 I found on the internet, this booted and ran tight settings twcl6 for me

 

Test 5ghz CPU 32m 1,4v mem auto Bios F6

 

First test I did was on Samsung to see if IMC does 2800, this worked.

Samsung result

2800 9-12-12-21 1,99v, loose rtl

 

Then I started with psc, using a kit for pretests 2400-2666 and trcd boot tests, on this imc of the Pentium showed big weakness, stable limit was around 2700 8-12-8, when I tested 2750 and 2800 the vga driver failed at win load, which was clear sign for imc or memory fail, with mems pretested for 2750 32m and higher, the blame is easy to put on imc, which failed same settings at other board as well. All results with all mem kits are obtained with memory and cpu on air cooling normal ambient temps, not all voltages are minimized

Due to Bios X04 not working with pentium K core oc from bios, I raised multiplicator with software.

 

PSC results

2667 8-12-8-28 tight 32m 1,77v

2702 8-12-8-28 tight 1,81v 32m

Validation 2800 C8-12-8-29 1,89v, 1m worked, 32m failed

http://valid.canardpc.com/zh32s1

 

BBSE result, 2600 worked, 2667 failed boot at any voltage, might have been to lack of skills or imc power but kit ran 2667 tight on various boards and cpu at z77 and z97 on other boards

2600 8-11-7-22 tight 1,88v

 

Elpida Hyper results

2000 7-7-6-20 1,58v not minmized

2200 7-7-6-20 1,75v

 

Generally, the board did a good job and I had lots of fun. Results shown were obtained with only a few hours of testing and a random CPU with not too good imc. I also tested some of the profiles, and I can´t recommend to use these, the samsung kit I used failed both profiles (2666 Samsung profiles) in spite of doing 2800c9 with hand dial and above 2800c9 on other boards, the 2667 tight psc profile set the main timings to 8-11!-8-28 twcl, the trcd of 11 on air for psc might be doable with handpicked vendor kits but for sure not with retail rejects vendors sell^^. The 2400-2600 PSC tight profile was not too bad and proved to be bootable and also might provide help for users without experience

 

Thanks for reading, I hope my small test was fun for you, all hardware used was retail, all results are on air, maybe it can give some support for normal ocers who want to use this board.

Edited January 3, 2015 by websmile

[suzuki]

This thread comes in a perfect timing,i have to receive my z97 Soc Force board this week,hopefully.

A small suggestion,put Asrock Timing Configurator instead of mem tweakit ,it's more easy to look for latencies .

Thanks for sharing and keep them coming,including tips on how to achieve those frequencies on psc & samsung.

[websmile]

Thanks, suzuki, I will do a few more tests with other cpu most likely and publish results using timing configurator^^ if I find the time

[dinos22]

those PSCs will bench 2800 6-10-6+ freeeeze em

[TASOS]

..... , I will do a few more tests with other cpu most likely and publish results using timing configurator^^ if I find the time

 

Many thanks for the input

[nachtfalke]

nice input!

 

still lot of work with this board tunning the bios for best mem performance , gigabyte from manny years haven´t a complete product...

Edited December 24, 2014 by nachtfalke

[websmile]

Yes, it might be my OS or I miss a setting, but I have impression that board is slower on 32m at same settings than asrock and asus, but you never know^^

Now with Timing configurator, 4690K and different mem kit

[sabishiihito]

Man you gonna make me pull that board out the box and hook it back up lol

[Massman]

Pretty sweet testing!

[websmile]

Thanks, the board is better than its reputation at memory oc but Gigabyte still has room for improvement on Bios and compability, I didn´t regret buying it for the test session, in the end I liked it much more than I expected

[nXXo]

psc air on the soc z97 no problem...but under cold it's not the same story

[websmile]

Could be, I saw only 2666 c6 on cold and not too often. Shouldn´t matter too much for "normal" user and ocer though, only very few people bench with mems on cold and there are lots of alternatives.

I made a last speed test at 4ghz before I unbuild system, no OS optimizations, board is at least competetive for 2d but still I am a bit slower than I was on last tests by competitor, save settings on mem and voltages

[Guest Bullant]

Nice Web,good to see your still enjoying your 32m Pi

[Mahameru]

Nice thread Michael, glad to see you play with Gigabyte

[websmile]

Thanks, Bull and Ivan - in fact I expected to suffer a lot, but honestly this was positive surprise and I had fun

[Jimba]

very nice results!!

[Farjam]

Now i'm working on it Michael (with that 3c6t) lol

[websmile]

Sounds as if you use AMD-like cpu on socket 1150 - you still have two threads more than me

[phobosq]

Last week I played with Hypers (A-Data 2200 and Dominators 1866C8) with 4690k on Z97OCF, both made 2200 7-7-7-20 1.65 easy, but 2400 was no go, no matter what voltage/timings - the most I did was 1M stable, 32M froze instantly. I tried 8-8-8, 9-9-9, 9-10-9 in combination with voltages from 1.65 to 1.85.

[websmile]

I have just my random Corsair left, but might try these in near time, on Z87 OCM they ran 1200 c8 around 1,73v iirc 32m stable, no idea if they can do same on GA - lots of Hypers fight with tight tertiaries and rtls on new systems and high frequency can have same effect, at the moment I tested some multi core benches on the board, it does quite good

Edited January 28, 2015 by websmile

[phobosq]

I tried to run D9GTR today (CSX Diablo 2000) with no luck - I was able only to run them as DDR3-1333 using one of the XMP profiles but the timings were terrible. I guess it makes no sense running D9GTR on Haswell