marsey99 AMD Overclocking Log

tinytomlogan

The Guvnor
Staff member
Hardware:

AMD X4 970 CPU

Biostar TA890FXE Motherboard

Scythe Susanoo CPU/System Cooler

Kingston LoVo 1600MHz 4GB Ram

No changes or substitutions to hardware to be made.

Benches:

CPU-Z - Max Screenshot

CPU-Z - Max Stable (30mins prime 95)

PCMark 2004

PCMark 2005

PCMark Vantage

PiFast

SisoftSandra

SuperPi

SuperPi 32m

wPrime 1024m

wPrime 32m

F@H - New profiles must be made for the competition "AMD-marsey99" Results will be taken for the best single whole day.

1st place bench = 5 points

2nd place bench = 4 points

3rd place bench = 3 points

4th place bench = 2 points

5th place bench = 1 point

All benches must be validated on HWBot for the OC3D team and a secondary hardware shot with with the screenshot in the background must also be posted.

All participants must submit results for all benches.

Extra requirements can be added at any time.
 
as per the rules of the internets any thread is pointless without pictures, so this is where i will be starting, with a picture or 2. these are just a few of the shots of the hardware Tom named in the op i have taken, if anybody wants a higher res version let me know and i will upload it for you too
smile.gif


so here they are along with some initial thoughts.

84231148893943861130.jpg


69700075974249848337.jpg


motherboard, the nice red, white and black scheme will please some football fans more than others, me i like it. enough room around the cpu sockets to fit any heatsink but your choice of orientation might be limited. im not sure why all the pci e x16 slots have been put so close to tell you the truth as that will limit the options for 4 way xfire but the option is still there none the less.

28414413337683926858.jpg


83197452247266864569.jpg


81905843357361831095.jpg


stills just dont do these sticks justice, the etching on them is like on the grey genesis range and the way it sparkles and shines is something you do need to see to believe. whats even better is that they heatsinks on them are not over sized so the height of these will not cause any clearance issues with your heatsink.

12620314551810829929.jpg


66971026033465272374.jpg


94935853925055091449.jpg


24893332172286298754.jpg


54249308643346725547.jpg


10488434775632553348.jpg


the susanoo. the back plate is a solid one which has all the holes to fit all current sockets, what more can you say. it and the screws are held in place by 4 rubber grommets. the base is absolutely beautiful, its flat and with an almost mirror finish which puts all the others i have seen to shame to tell you the truth.

20059338221326493182.jpg


74943480192811935040.jpg


48598684392571056644.jpg


the fans. 4 100mm ones with a control unit on a pci expansion bracket. at full speed they are not as loud as you would expect as they only spin up to 2k rpm and on low they are silent at 700 rpm.

15466827938132884857.jpg


77328206572499675588.jpg


enough tim supplied for about 3 mounts, i always do 2 anyway to check if what i had used spread out enough and then do it again with either less or more depending on how the first was.

11383079054410412255.jpg


89663256791050291978.jpg


82085549671070703120.jpg


now this is the only part about this heatsink i had an issue, having to fix it in place upsidedown. please note this is only the amd mounting kit and not for the intel one as there are thumb screws in the box for those sockets. needing to screw it in place from the backplate made it a tricky job as i needed to try and line up the board with the sink when the socket is surrounded but heatsinks and ram made this very tricky. a big plus point goes to the stand off that come with the heatsink tho, as you can see in that last pic there are push pin ones included which will make fitting the board into a case with the heatsink attached a breeze.

95779972123938559070.jpg


93017495054903485288.jpg


94897113816101483170.jpg


coming back to the fans, as you can see you get 3 clips in 2 sizes. the 2 shorter ones go along the bigger array of fins while the 1 longer on goes on the side with the smaller array and once all 3 are clipped in the fans are very secure.

12177896733394182155.jpg


79780988697919045380.jpg


52122833772963565241.jpg


once in place it is tight for room as you can see but nothing is touching and everything is in place. the last picture there shows best how much room under the heatsink you get, even with the aftermarket heatsink on my gtx460 (not the biggest of cards keep in mind) there is still room and it has heatpipes which stick out around 5mm over the top of the pcb.

38112605985148141150.jpg


32481582960774314184.jpg


67737751334412394821.jpg


just to round out the rest of the bits i will be using there is a gigabyte gtx460 oc, the corsair hx650w psu, a samsung 250gb spinpoint hdd and a liteoen dvd rom.

18799775639918124919.jpg


there we go. all together and sat on its box awaiting to be punished
biggrin.gif


edit

1st go.

well i wasnt playing on posting anything so soon as i wasnt sure i would have chance (and i only have my phone atm...)but whilst seeing what was new on redtube i thought this was a better use of my time.

1823455.png


x35sso.jpg


http://hwbot.org/community/submissi...pu_z_phenom_ii_x4_970_be_4500.01_mhz?new=true

i1z39u.jpg


i was interested to see how far it would go with 1.5vcore while just making use of the unlocked multi on the be
smile.gif


pic is crap because its from my phone, but thats something im working on
wink.gif


more to come soon
biggrin.gif
 
time was the factor tbh as i only had an hour, i just wondered how high it would go on 1.5v and while 4.6ghz would post it was locking up but im sure with a little more love its going to happen.

it was pretty warm before too at a little under 23c but its meant to get a little chilly later in the week so even higher might be on the cards
biggrin.gif


edit

a little cooler now but its getting even colder the next few days
biggrin.gif


2dawpib.jpg


http://hwbot.org/community/submission/2170148_marsey99_cpu_z_phenom_ii_x4_970_be_4623.09_mhz

now im off to bad
biggrin.gif
 
Whilstwaiting for the air temp to come down last night I decided I would see how far the ram would go on stock timings. Normally this would be my first port of call when starting to oc but the unlocked multi was too tempting not to play with first.

So I dropped the multi on the CPU/NB, set the ram to stock timings and 1.6v and went about upping the fsb. The ram went to 1880 without a hitch but when I restarted it to try it higher nothing, no bleeps, no code nothing. Its like the reset but was just being held as everything is firing up but its not starting to post.

I stripped it down and have left it powered down over night hoping the gods of ocing or the hardware fairy's would bring it back to life. I'm about to try it again now so send me your positive thoughts peoples
smile.gif
 
well it worked
smile.gif
thanks for the good vibes
biggrin.gif


after it throwing a tantrum last night and clearing the cmos i figured i would just start again armed with the knowledge i had picked up so far. after the bios had been cleared i loaded the opti defaults and set the ram to spec, saved that as a profile in the bios and f10 return. after it had restarted i then turned off all the options not needed for benching like the sound and nic, firewire, ide controller, power saving and so on i then saved it as a new profile called bench.stock so i had a nice clean profile to start from if the worst happened again :|

so knowing that the ram would hit 1850mhz+ i thought i would continue where i left off last night and test multi vs fsb but with keeping the ram under that speed with an easy 4ghz oc to keep it interesting
smile.gif
starting at the stock multi and dropping it down as i increase the fsb.

228x17.5

85546387433786140322.jpg


235x17

87744724239459268710.jpg


242x16.5

88553751903634829778.jpg


250x16

71714118059255422747.jpg


258x15.5

48130010788990330935.jpg


as you can see the times come down as the fsb goes up by around half a tenth each drop on the multi, even as i change the ram ratio toslow it down and keep it happy. now 266 would of been the next picture but it wasnt booting on the settings i used but was close. now given it was the middle of the day when i was doing this and the air temp was starting to rise and after the board spat its dummy last night i bottled out on giving it the full beans with the juice so i could see what i could get out of it around 258.

260 is solid as a rock all the way to x17 and only hitting 45c on the cores fully loaded i know i have the cooling capacity for more juice, even in the middle of the day but as i was running out of time to come here and post this i aired on the side of caution. i only ran super pi 1m in the hope that i can get a full night of tweaking in tonight and give the guys something to aim for as it feels like im playing catch up atm.

91769341678717602598.jpg


<click it to see as its a big un
wink.gif


http://hwbot.org/community/submission/2170671_marsey99_superpi_phenom_ii_x4_970_be_15sec_656ms
 
so after some trail and error it seems i cant get over 4.5ghz with all 4 cores avtive, so for these few runs they are all @4.42ghz with the ram @1732mhz 9-9-8-26.

02440127120454259792.jpg


http://hwbot.org/community/submission/2171884_marsey99_pifast_phenom_ii_x4_970_be_24.88_sec

pifast, 24.88s

50178297658532828340.jpg


http://hwbot.org/community/submission/2171887_marsey99_wprime_32m_phenom_ii_x4_970_be_8sec_313ms

wPrime 32m, 8.313s

79258627195835178757.jpg


http://hwbot.org/community/submissi...me_1024m_phenom_ii_x4_970_be_4min_25sec_219ms

wPrime 1024m 4:25.219

not as fast as i was hoping tbh but for the clock speed they aint bad, i will be trying to push them harder when i get a new burner as mine died the night i got these bits and i only have an old rom atm so im stuck with the os i already have on disk.
 
Wow there really isn't much between the chips! I'd expect there to be more of a difference I guess. Great work Marsey!
 
Wow there really isn't much between the chips! I'd expect there to be more of a difference I guess. Great work Marsey!

cheers todd, looking at the other guys results so far i think i got the runt of the cpu litter tbh as i cant get it stable around the same speeds as the others on the same volts but it seems like strawbs really drew the short straw.

ram on the other hand i think i did better, well that or im giving mine much more voltage xD
 
so last night while it was cool i figureed it would be a good way to get my p95 run done and after quite a few goes i finally got it to do the whle 30 minutes at a whopping 4.2ghz :|

97232090550287969722.jpg


38002160878269349661.jpg


unsure.gif
tbh i had a feeling this cpu wasnt as good as the others after how high they could go with 4 active cores but i guess this is the giveaway, 1.55v for 4.2ghz
unsure.gif


as you can see the dslr has gone back to my old man, well its got packed away as its going to the isle of man over night if the ferry isnt canceled so im down to talking pics with my phone :|

this afternoon i had a couple of hours free so i thought i would try and push the clocks in a different way with less nb/ht speed but higher ram mhz and cpu multi. now it topped out shy of 4.5ghz again but did gain me 60mhz @ 4.485ghz with the ram now at 1840mhz vs the 1732mhz it was. not i have only done a quick maxmem run to see but i have lost abit on that but managed to get these from the cpu.

19779710529210100951.jpg


15min 49sec 578ms - SuperPi 32m

http://hwbot.org/community/submissi...henom_ii_x4_970_be_15min_49sec_578ms?new=true

11720777186673817451.jpg


15sec 547ms - SuperPi 1m

http://hwbot.org/community/submissi...erpi_phenom_ii_x4_970_be_15sec_547ms?new=true

86815014090456518119.jpg


28.45 sec - PiFast

http://hwbot.org/community/submission/2172189_marsey99_pifast_phenom_ii_x4_970_be_28.45_sec?new=true

which are all just better than my previous scores, but here come a problem. i just plugged my phone in to upload the pictures of these to be told that thre might be a problem with some files on this device and if i want windows to scan and fix the issue to stop future issues, which is and has been going for ages now and has yet to finish. the pictures from this afternoon are also gone
sad.gif


now my plan for later is to test if the ram will go any tighter at this speed which will help me beat these scores again, and allow me to get pictures of them but for now im waiting to see if this scan finds the pictures or kils my micro sd card :/
 
Good job on hitting 1840, probably going to be your limit but you can take CL and tRP down a notch I should think
smile.gif


NB speed will help a lot also.
 
I just got home now so I'm going to see which eps of family guy are on tonight and either test the timings after they go off or instead of watching it
biggrin.gif
 
Yea they was crap ones on tonight so is did just turn it off xD

Maxmem ran once then started locking up so I reran the settings in aida to get some comparisons and now the boards not posting again....

Only thing I have noticed that can stop it from recovering is when you push the ram too far as its great sorting itself from a bad CPU oc.

Its sat clearing the cmos as I speak in the hope it comes back to life before I get too far down this bottle
laugh.gif
 
this afternoon the board came back to life, so i figuered i would go easy (ish) on it and start at the bottom and work my way up
wink.gif


taking the timings from the xmp 1.35v profiles on the ram i wanted to see how far i could push each set on a higher voltage before i needed to relax the timings to the next settings. as you can see in this picture these are the timings i used untill it was unstable (by which i mean wouldnt post/boot/do anything in windows without crashing, not memtest86 overnight stable
wink.gif
) at which point i would use try the next set up starting with the subtimings, then the main before increasing the cas latency.

87501648813241758755.jpg


cas6

22361493198680080367.jpg


10448518255689444116.jpg


16118754863154396689.jpg


27547472376082795828.jpg


43522309860838108121.jpg


92253841455112658230.jpg


47743033858364793302.jpg


74867300442694621060.jpg


09933562302350083226.jpg


notes.

1307 i had to let the subtimings go to the next set.

1360 i had to up rcd and rp

then i hit the wall at 260 and cas 6 would not post on the next ratio.

cas7

54046825170706643759.jpg


61263191869348611360.jpg


18715717367484301101.jpg


15526798635625712027.jpg


60010731673700343249.jpg


86037561817144898403.jpg


29981964078877509235.jpg


68008418852553690434.jpg


57074612319740878622.jpg


15656947405566500644.jpg


25043587470647992538.jpg


93049475329424407602.jpg


notes.

1500 i had to up the subtimings.

1533 i had to up the tcd.

1567 i had to up the rp

1700 locked up after taking that screenshot.

260 would not post with these settings nor would it post at slower ram speeds with it cas 7 on the next ratio.

cas8

11415608600841937169.jpg


64550286553753364166.jpg


72694059378022381297.jpg


notes.

1840 i needed to knock the subtimings and the rcd up.

if you have been reading this you will of seen that 4.5ghz is around my limit for 4 cores so with that in mind i kept dropping the multi as i upped the fsb to test the ram and eliminate the cpu from the equation (but i have found that its much more stable on some ratios than others when pushing the fsb, more on this after more tests i think) and to try and keep the results as much about the ram as i could i tried to keep it as close to 4.5ghz as i could.

for these tests all voltages where the same at;

vcore - 1.55v

vdimm - 1.7v

cpu>nb - +150mv

nb/sb/ht - +50mv

now here we are again at 1840, for the life of me i dont know what i changed the night i saw 1880 from these sticks but it was also the same night i had 4.7ghz posting but failing to boot. now it could of been the bios revision or the temps or just that the ic was shiney and new and has since degraded idk but i seem to keep hitting a wall here, much like the one im hitting at 260 fsb.

the more i tried the less it wanted to play but im pretty sure its the nb thats holding me back here now, i am also pretty sure that on this voltage and cl 9 these sticks will sail past 2000 mhz in the right board/cpu/nb combo but as it is its only a theory and one i cant test, but i know ian can at a later date in his p67
biggrin.gif


now for those of you with a tl:dr mindset
wink.gif


cas6 - 1333mhz

cas7 - 1700mhz

cas8 - 1840mhz (maybe higher?idk as i cant test it)
 
armed with the info i got from the ram tests i thought i would see if i could get my scores better with the new ram settings over my older ones.

87062731661896274674.jpg


77964174289031142449.jpg


SuperPi - 15sec 265ms http://hwbot.org/community/submission/2172653_marsey99_superpi_phenom_ii_x4_970_be_15sec_265ms

83080667916731862504.jpg


91933821636584646878.jpg


SuperPi 32m - 15min 27sec 656ms http://hwbot.org/community/submission/2172654_

this took some doing to tell you the truth but bot am i happy with it, not only the fastest time on air but i beat 2 guys with cascades and i didnt do wazza lol so i have more to get from this 1 yet, i was just running out of time so i left it here for now.

36156854021069519073.jpg


29893911168298365766.jpg


PCMark 2005 - 12608 marks http://hwbot.org/community/submission/2172655_

not a great score tbh but it is the 1st time i have got this bugger to run, found out that wmp11 will make it work even tho it had and asked for wmp10 :\

still cant get the grammar test on 04 to play ball and as you can see from the pictures i spent a good while trying as it was dark when i finished with super pi and its day light on pcmark
ohmy.gif


sorry about the quality, or lack of, in the pictures but its from a phone :|
 
Back
Top