Lets take a look at Gigabyte's latest X58A-OC

[chew*]

This board arrived earlier today so lets take a look at it put it through it's paces and compare it to another Gigabyte offering in the suggested price range.

I'm not going to bore you with writing a book on every single component used, instead I will focus on real world in the form of OC's and results.

For testing purposes I will be ditching the gulftown cpu as imho chip is always the limiting factor. I will be opting for the still very popular 920 DO ( I might add my own personal chip ) as that is a chip that quite honestly a board can make or break the overlockability.

This board in my opinion is all about bang for buck, all the stuff you want and for the most part none of the stuff an overclocker doesn't need.

The box art is very simple and to be honest it appears that Gigabyte managed to save endusers money by going sparse on the advertising and packaging.





I zoomed in on this one, I think this may the first time marketing has nailed something on the head accurately. No charts showing how the caps are better just simple and straight to the point, we used them to make insulating easier.



As I said they kept the packaging very minimal.





Very simple yet complete bundling, I might add there are 6 plugs included for measuring voltage where other competitors offer 6 pinouts but only 2 plugs?



Another shot of the board. I was really hoping they would move the clear cmos and pwm frequency switch to a more remote location in the final rev.



Southbridge has a thermal pad and very good contact.



PWM exposed.



Northbridge also had good contact.



What happened here and here though? Looks like poor contact for mosfets, hopefully this will be a non issue when I replace the pushpins which i have never been fond of.





Last but not least the board I have decided to compare it to through benchmarks, general useage, operation and features.

This was sort of a tough choice and in the end boiled down to suggested pricing.

[chew*]

First impressions as I am playing with this board a little checking out the bios, confirming the 4G button operation etc etc, getting a feel for her.

So my first impression is.........Where the hell is my IDE , for me IDE high pci and ocing go hand in hand especially in 3d.

This is definitely a con in my book especially considering the X58A-UD5 which I have chosen for comparison does have my beloved IDE.

Dual ps2 is a plus. Bios is simple and familiar, no surprises so far.

4G button worked fine at least to bios no OS so I didn't try booting it.

Here's a quick shot of what I threw together to start with, Fluke is monitoring vtt for now, side note with 4g button it went up to 1.44vtt.



Ok so after warming up to this board for a little while I think I have the hang of it for bloomfield clocking.

The good

I have never had this chip over 234 bclk before on a gigabyte board and that was with DI, its booting 240 on air now.

The bad

QPI slow mode and dual channel for the moment.



You tube of the board booting 240 bclk normal ambient air.

http://www.youtube.com/watch?v=hI2DJy3OGFM

I ran this before I started getting nitty gritty just in case anyone was interested, Seems that the board is good for 220 bclk for crunching or daily use. Higher refused to pass no matter if ram cpu or uncore speeds were dropped.



After that I pounded on the board for a while till I found the effeciency I was looking for.

Made a goal, top 20 32m for 920 class or bust. Goal achieved albeit barely, it's also the lowest clocked in the top 20 32m 920 class

I think It's safe to say that the board is not a slouch in 32m effeciency at least with bloomfield.

Hopefully this chip scales at temps beyond phase change and i can run a slightly more tuned run



Ok conformalled just the general socket area from dimms to NB to PWM ( enough for DI ) and gonna see if this chip decides to scale any more with some DI seeing as thats all thats in my budget for the time being



Alrighty so my main focus was 32m, I ran a quick pifast to just to get a ballpark how the board stacks up to others.

Both results are where I would hope they would be after browsing the hwbot rankings for similar speeds.

Unfortunately my chip isn't the greatest but the board made up for it in 32m effeciency enough to grab a top 5 spot in the core i7 920 32m class.

Just look at that uncore





One thing to note which is peculiar is when I tend to get to frisky with voltages at least with bloomfield I keep seeing this error message



Other side notes as I think of them.

Vtt overvolts by .050-.060 ( measured ) not really a big deal as long as you are aware that say 1.54 = real 1.6.

Exceeding PCI over 120 drops my old raptor drive which is probably the best high PCI sata drive I own, symptom over 120 PCI E no boot device detected after post screen, ergo my complaint about no IDE on this board.

Ok found more time lets go over some other things on this board.

As always on Gigabyte boards be sure to CTRL+F1 to unlock hidden options, one option of importance for most overclockers that is hidden here is spread spectrum which by default is enabled.



Voltage increment options

Vcore Range 0.80-2.12


Vtt range 1.12-2.0


Pll range 1.30-2.52


I would really like to see finer increments for the lower range, you can see here the upper range has finer increments which doesn't do most overclockers any good when sub zero.


Dram


ICH


IOH


I grabbed another chip brand new out of the box that I can gurantee by stepping without testing is less than stellar ( CO 940 ).



I gave that a quick shot with bclk on phase as well, for those curious it can bench 4600 cpu/4600 uncore max on phase.

3d testing bclk without slow mode.
http://www.youtube.com/watch?v=vBUKoi4gGu8

32m testing bclk with slow mode


248 seems to be a patternable limit at least for 32m, I can probably run 1m at 250, been booting 250 without to much work.

Things to note, board seems to favor 112 PCI, higher is only needed to surpass 248 bclk. With this CO chip higher PLL seems to aid bclk dropping it will hang at splash screen ( I think i only went up to 1.88 )

Broke through the 248 wall by 1 bclk, Ive already broken it before but effeciency was bad. This time it's working right.

Here's that C0/C1 940 chip........scaling on DI was horrid, as was a wicked bad IMC CB that I had to work around.

#4 in it's hwclass at repectively less clocks.



Short you tube I made near the end of the run.

http://www.youtube.com/watch?feature...&v=g12hXEptVOY

Yet another 32m run on another chip that I just opened brand new out of the box.
I will give this a shot on DI tommorow and after that my 32m pi fetish is over and we can move on to some other stuff.

#3 core i7 930 class for now


Ok fetish over, first place 32m for core I7 930 class, certainly not my best work and its rather ugly but the chip was a nightmare to work with on DI so it is what it is.



Had some DI left over so I figured i'd test the board out under load, my other 920 will max out bclk in 3d, this chip not so good it has a frequency limit.

Nothing tuned really only had enough for one run.

[chew*]

Bios used for the testing hosted by hwbot, courtesy of a bios dump directly from my board.

http://91.121.148.119/downloads/x58a-oc/X58AOC.014

Time for a direct features compare from my point of view as an overclocker ( least I think I am in my mind )

First up socket area.

X58A-OC 1
X58A-UD5 0

Scoring based on if your an eraser type guy it sucks to be you insulating the UD5 as the socket area is very very busy.

OC


UD5


Next up cool buttons and switches.

X58A-OC 1
X58A-UD5 0

Scoring based on the fact that you have to use ET6 software if you want on the fly overclocking with UD5

OC


UD5


Measure points

X58A-OC 1
X58A-UD5 0

Scoring based on me being lazy and not having to solder in measure points on OC board versus the UD5.

OC picture only since UD5 is N/A


Port 80 debug

X58A-OC 1
X58A-UD5 1

Both boards have port 80's which is a big + in my book.

OC


UD5


Clear CMos acessibility and ease of use

X58A-OC 0
X58A-UD5 1

I do not like location of the Cmos clear at all on OC board, it's a condensation issue waiting to happen and if it doesn't cause an issue while benching the switches internals will most certainly corrode over time if moisture gets in it.

OC


UD5


Hard drive ports.

X58A-OC 0
X58A-UD5 1

The UD5 wins this one for me not becasue it has MOAR sata ports but due to the fact that it also includes IDE.

OC


UD5


UD5


Dual bios

X58A-OC 1
X58A-UD5 0

OC board at the flick of a switch where as to boot off backup bios you need to short pins on UD5

OC


UD5


Input power, not going in to specifics about input power specs whatever but 8 power/ground feeds makes me feel safer that i'm not melting wires on my PSU.

X58A-OC 1
X58A-UD5 0

OC


UD5


PCI layout

X58A-OC 1
X58A-UD5 0

This is a no brainer, One board has full 4 way crossfire support ( dual slot ) and one doesn't.
Neither board supports 4way sli.

OC



Clears ram fine


UD5




Back panel I/O

Me I am impartial here I call it a tie since both have Dual PS2.

X58A-OC 1
X58A-UD5 1

OC


UD5



Power input for PCI lanes.

X58A-OC 1
X58A-UD5 0

This is easy, UD5 doesn't have them

OC



Now i'm not gonna score based on fan headers......

OC has 7 UD5 has 6, both have plenty.

OC





UD5





Both boards use an ISL chip and the SiC mosfet drivers.

OC


UD5


OC


UD5


Last but not least we have this switch that changes switching frequency, haven't read up on it yet so haven't used it.



Final tally from an overclocker's point of view.

OC 9
UD5 4

Performance comparison OC versus UD5 head to head.

Testing methods consist of the following.

All benches run 3 times back to back then averaged.

Fresh boot before running the next different benchmark

OS made prior on a master drive then duplicated to bench drive to ensure a fresh OS for both platforms.

Ht shut off in benches not benefitting HT, cores set to priority in benches not utiliyzing all 4 cores.

Here's a shot of the drive duplicator. This is a hardware copy not an image and is very good at keeping things consistant.



Settings for OC board.









I would like to take this time to thank a fellow member on XS (Ron aka Runmc) for making me a kick ass phase change that makes my life alot easier.



The OC board in operation, the fluke is monitoring vtt.



OS and settings I have chosen for the compare.

Vga will be at stock for consistency.



Settings for UD5, *note* I had to run 1.40 vtt to match the OC board but also note the OC board overshoots vtt.







The UD5 board in operation



OS and settings I have chosen for the compare.

Vga will be at stock for consistency.




Included along with final results will be middle results from all the runs.

OC 3D 11 = 4940 1
UD5 3D 11 = 4939 0

OC


UD5



OC 3D vantage = 22035 1
UD5 3D vantage = 21988 0

OC


UD5


OC 3D 06 = 28257 0
UD5 3D 06 = 28359 1

OC


UD5


OC 3D 05 = 37092 0
UD5 3D 05 = 37162 1

OC


UD5


OC 3D 03 = 90914 0
UD5 3D 03 = 90969 1

OC


UD5


OC AM3 = 239,784 0
UD5 AM3 = 241,337 1

OC


UD5


OC Heaven = 795.8148 1
UD5 Heaven = 795.7638 0

OC


UD5


OC Maxxmem = 1609 0
UD5 Maxxmem = 1611 1

OC


UD5


OC pifast = 17.92 0
UD5 pifast = 17.88 1

OC


UD5


OC spi 1m = 8.601 1
UD5 spi 1m = 8.611 0

OC


UD5


OC spi 32m = 7m 44.517 1
UD5 spi 32m = 7m 44.580 0

OC


UD5


OC wprime 32m = 5.39 0
UD5 wprime 32m = 5.33 1

OC


UD5


Score

OC = 5
UD5 = 7

Now what does this mean? It could mean alot of things.

UD5 is more tuned for bloomfield?

OC board is tuned slightly slacker to hit higher overall speeds?

It simply could mean the bios on UD5 is more mature.

Regardless none of the results are far off from each other.

For anyone interested in the complete 3 runs I did per board should they wish to analyze them they can be found in the following links.

X58A-OC http://chew.ln2cooling.com/?Qwd=./OC&Qiv=thumbs&Qis=M

X58A-UD5 http://chew.ln2cooling.com/?Qwd=./UD5&Qiv=thumbs&Qis=M

[Hondacity]

best pics of the board

those tantalum caps are yummy...

bclk to 265?

is that retail packaging? cardboard box? lol

[chew*]

best pics of the board

those tantalum caps are yummy...

bclk to 265?

is that retail packaging? cardboard box? lol

Yes this is supposed to be full retail packaging. I like it if they saved on bling but made the board sing I'm all for it.

Will try for some better quality pics in morning i'm shooting at night time right now and the orange/black contrast makes shooting this board kinda hard.

[Vicey]

Great pics. I love the look of the build with that orange flare. Finding things to match that colour scheme would be difficult at-least not without custom paint jobs.

EDIT:// Match made in heaven? http://i.imgur.com/DOsvn.jpg :P

[chew*]

Great pics. I love the look of the build with that orange flare. Finding things to match that colour scheme would be difficult at-least not without custom paint jobs.

EDIT:// Match made in heaven? http://i.imgur.com/DOsvn.jpg :P

I think some people will either hate or love the color scheme.

It might be challenging for a modder who like's to color code component wise say VGA and memory but i'm sure it can be done.

It would certainly be a bright build.

[Tiltevros]

nice board looking forword for some benchies

[Raja@ASUS]

Nice work so far Chew, at least you are able to keep an objective frame of mind in light of receiving a sample. This at least has the makings of a review instead of an advert.

[dinos22]

whats with air cooling bro

[sin0822]

yea they really did save some blink with the box. Nice job!

[chew*]

whats with air cooling bro

http://www.xtremesystems.org/forums/...postcount=3213

Not sure if i can afford ln2 till I get my car situation straightened out, probably won't need it for this chip though, phase in a bit just feeling the board out.

I may have air cooling but I found CAS 7 faster than you.

[dinos22]

oh dear

[bassplayer]

Were you hitting a hard wall at 240 on dice too? My UD9 did 257 (never tried higher) and my RIIIE did 266. Since I'm using a 970, if I want to bench at 6GHz plus, I'm going to need 240+.

[chew*]

Were you hitting a hard wall at 240 on dice too? My UD9 did 257 (never tried higher) and my RIIIE did 266. Since I'm using a 970, if I want to bench at 6GHz plus, I'm going to need 240+.

Had it up to 248 on phase today but my temps were 11C

Haven't gone colder yet.

[Raja@ASUS]

If that's 248 with QPI fast mode intact it's good.

[K404]

Check you out, Chew, with all the best toys


Looking forward to the ragged edge and some Uncore abuse

[Biker]

Good honest reporting .... now where is the "like" button

I do not like the lack of an IDE port either or the very poor heatsink contact around the cpu but at least this is now flagged up and will be checked / modded when I get a few in for builds....

[chew*]

If that's 248 with QPI fast mode intact it's good.

No sir, everything is with slow mode.

I'm not sure if it's this chip but so far every board I have had it in 235 is the limit for normal QPI.

Wasn't benchable mind you that was just the max boot limit.

[Raja@ASUS]

Still not too bad overall Chew. Most of the non NF200 boards seem to top out south of 240MHz BCLK with fast mode for 3D benchmarking, so this is competitive.

[Olivon]

Nice Brian and great Bclk

That's suxxx time for your car though

[chew*]

Alot of stuff updated in second post.

[Hondacity]

uncore woooooo ...dice temps for 5.2g uncore?

vdimm adjustments...i was hoping 10mv or 5mv

[Buckeye]

Very nice Brian, thanks for the review of what this board can do.
920 bclking is still a lot of fun

[chew*]

uncore woooooo ...dice temps for 5.2g uncore?

vdimm adjustments...i was hoping 10mv or 5mv

Cpu was reporting -45 ish, pot was reading -70.

Chip is known good for -130, might be more in it but I fear i'm at a bclk limit at least for 32m.