Titan Goliath at 3.8 GHz

mhk1058

mhk1058

mhk1058

Associate
Joined
14 Dec 2009
Posts
130
If anyone is interested in pushing their Titan Goliath a bit further I have mine running at 3.8GHz (Well, 3.78) very easily with just a couple of changes.
Please be aware that changing anything from the standard 3.4GHz as supplied means you say cheerio to your warranty and is entirely at your own risk.

Here is a list of the main settings changes from what would be mobo standard to achieve the 3.4 goliath and then the changes I made to get to 3.78. Remember, all i7's are not equal, mine is the D0 revision, earlier models will need more juice to the Vcore, maybe other settings changes and may get hotter as a result.

First off, enter BIOS and go to Advanced CPU Features. Enable 'Turbo Boost', this gives you a multiplier of 21. I couldn't get things stable with 20. NB It will still say 20 in the BIOS, use CPUID, CoreTemp, etc to see x21


Turbo Boost
OCUK 3.4 Disabled
My 3.8 Enabled to give x21
BCLK Control
OCUK 3.4 Enabled
My 3.8 Enabled (of course;))
BCLK Frequency
OCUK 3.4 170
My 3.8 180 (21x180= 3.78GHz)
PCI Frequency
OCUK 3.4 101 (?)
My 3.8 101
CPU Clock Drive
OCUK 3.4 900mV
My 3.8 800mV
PCI Express Clock Drive
OCUK 3.4 900mV
My 3.8 900mV
Performance enhance
OCUK 3.4 Standard
My 3.8 Standard
System Memory Multiplier
8 for both
Timings
OCUK 3.4 9,9,9,28 (?)
My 3.8 9,9,9,24
DRAM Timing Selectable
Quick for both
QPI Link and Uncore
Auto for both
Load Line Calibration
Enabled for both
CPU Vcore
OCUK 3.4 1.30000V
My 3.8 1.28125V (Yes.... lower!)
IOH Core
1.22V for both
All other voltages AUTO

So there you have it. Some of the original settings were a bit questionable, like 101 PCI Frequency and memory timings, I would be interested to know if other Goliaths had the same from stock.

I have run OCCT for 2 hours, Prime95 for 4 hours and all seems fine. Temps just touch 80C occasionally while running these tests. Putting fans on high brings them down 1-2 degrees. Real world use I'll be keeping the fans on low and temps will be well under 70C. Careful if you have the basic CoolerMaster Case, cooling may not be as efficient as the Antec 300. Always use RealTemp or CoreTemp to keep an eye on things

Interestingly, one more notch to 181 and system crashes in PassMark. Probably could sort it but very happy where I am.

Once again, you can break stuff doing this. If you do, OCUK, quite understandably, won't want to know.

Edit:

OK, couldn't resist actually seeing 3.8GHZ

Settings now that differ from OCUK Goliath:
Turboboost Enabled (x21)
BCLK Frequency 181 (181x21= 3.8GHz)
CPU Vcore DOWN to 1.23125
Also set memory back to 9,9,9,28 which is how it was set originally

Temps peak at 77C in OCCT test (Core #0= 77 & 4 degrees higher than rest) and 76C in Linx (Core #0 3 degrees higher than rest)
Linx throughput peak 46.77GFlops

That's it, chuffed to bits, Vanqish performance for Goliath price.
 
Last edited:
Those temps are better than I was getting with a Megahelam on less voltage, are you sure it's not throttling under load? sweet chip if not.

Try using Linx with all your memery, it also gives you a performance rating to ensure it's not throttling, you should be getting 45-50gflops.
 
mmj_uk said:
Those temps are better than I was getting with a Megahelam on less voltage, are you sure it's not throttling under load? sweet chip if not.

Try using Linx with all your memery, it also gives you a performance rating to ensure it's not throttling, you should be getting 45-50gflops.
Click to expand...

Thanks for the tip, not sure how to post a screenshot here without an external host or even if you can but LinX showing 46-46.5 gflops - hopefully that's in the right ballpark :p
 
Liam-89 said:
my titan goliath runs at 86-86-81-81 at 3.40ghz , no way i could handle 3.8
Click to expand...
That seems a little hot - what case do you have, what's your Vcore and are things getting dusty inside?

If you have the D0 revision, I'd be tempted to get the cooler off and give the base a polish and re-seat it.

My temps are a little higher now. As delivered from OCUK, no tinkering at all, all cores were low 70's at 100% in Prime95
 
mhk1058 said:
That seems a little hot - what case do you have, what's your Vcore and are things getting dusty inside?

If you have the D0 revision, I'd be tempted to get the cooler off and give the base a polish and re-seat it.

My temps are a little higher now. As delivered from OCUK, no tinkering at all, all cores were low 70's at 100% in Prime95
Click to expand...


antec 1200 , all fans at medium including the front fans turned up abit. not 1 bit of dust at all

vcore is 1.2875 or something 1.264 idle ,basically w/e the overclocker staff set it to. My temps idle are 44 42 41 39 , thats my room window open tho and its pretty cold
 
Last edited:
Liam-89 said:
antec 1200 , all fans at medium including the front fans turned up abit. not 1 bit of dust at all

vcore is 1.2875 or something 1.264 idle ,basically w/e the overclocker staff set it to. My temps idle are 44 42 41 39 , thats my room window open tho and its pretty cold
Click to expand...

That's pretty much what mines idling at now. There is quite a difference between chips, particularly if you have the earlier C0 revision?

If it's D0 I would say it does seem a bit hot if you have all the fans on etc. (But within safe range). The Antec 300 case just has a top and rear exhaust fan, putting them on high (from low) only seems to change things a degree or two - maybe different in the summer.

What are you pushing it with to get the high temps??
 
Last edited:
mhk1058 said:
That's pretty much what mines idling at now. There is quite a difference between chips, particularly if you have the earlier C0 revision?

If it's D0 I would say it does seem a bit hot if you have all the fans on etc. (But within safe range). The Antec 300 case just has a top and rear exhaust fan, putting them on high (from low) only seems to change things a degree or two - maybe different in the summer.

What are you pushing it with to get the high temps??
Click to expand...


prime95
 
liam thats not right mate not with that case...my sig runs max 82c under prime and that was in the summer time. Take the cpu cooler of clean it well and apply a very thin spread of TIM and reseat im sure your temps will go 70 odd after that...if not that cpu needs RMA mate
 
BigBruiserAl said:
liam thats not right mate not with that case...my sig runs max 82c under prime and that was in the summer time. Take the cpu cooler of clean it well and apply a very thin spread of TIM and reseat im sure your temps will go 70 odd after that...if not that cpu needs RMA mate
Click to expand...

ye well , i will phone on monday and find out what my options are.
 
New Settings

OK, couldn't resist actually seeing 3.8GHZ

Settings now that differ from OCUK Goliath:
Turboboost Enabled (x21)
BCLK Frequency 181 (181x21= 3.8GHz)
CPU Vcore DOWN to 1.23125
Also set memory back to 9,9,9,28 which is how it was set originally

Temps peak at 77C in OCCT test (Core #0= 77 & 4 degrees higher than rest) and 76C in Linx (Core #0 3 degrees higher than rest)
Linx throughput peak 46.77GFlops

capturezy.png


That's it, chuffed to bits, Vanqish performance for Goliath price.
 
Last edited:
is that with the akasa nero? , my temps are like 87 87 82 82 , and thats at 3.4ghz. Ocuk gonna send me more artic silver 5 to resit the heatsink see if that helps

mhk1058 , whats your temps idle?
 
That's too hot. Yes, mine is totally as supplied with the Akasa Nero and the 2 Antec 300 fans on low. Idle temps as below. Ambient is about 21-22 at a guess.

capture2q.png


It's likely to be just the cooler needs seating again, it's a big old thing and could easily move in transit.
 
Last edited:
Some chips do tend to run hotter than others, but I do agree those temps are pretty high for 3.4GHz.

The 2nd i7 chip that I got ran 15c hotter than the chip I had previously, at stock with a TRUE with 2 fans in push/pull it would idle at 51c, and under load it would reach the low 70's.

I tested 4 different HSF's on that chip, ( 2 x TRUE's, Zalman 9700NT and the stock cooler) and I got the same results with each heatsink, I wasted about 3/4 a tube of AS5 with all the testing I did with that chip!!

I got the chip replaced in the end thankfully!!
 
DavyBoy said:
Some chips do tend to run hotter than others, but I do agree those temps are pretty high for 3.4GHz.

The 2nd i7 chip that I got ran 15c hotter than the chip I had previously, at stock with a TRUE with 2 fans in push/pull it would idle at 51c, and under load it would reach the low 70's.

I tested 4 different HSF's on that chip, ( 2 x TRUE's, Zalman 9700NT and the stock cooler) and I got the same results with each heatsink, I wasted about 3/4 a tube of AS5 with all the testing I did with that chip!!

I got the chip replaced in the end thankfully!!
Click to expand...

Also check to see if you have C0 or D0 revision - C0 will probably run hotter
 
Who me?!?!

I've never owned a C0, and I would imagine if Liam purchased his system recently, then he will have D0 in his rig.
 
DavyBoy said:
I just assumed you were talking to me because you quoted my post...
Click to expand...

tempsl.png


im sure its a d0 , also these are idle temps which seem close mhk1058's . so why the hell does it leap up to 87-87-82-82

totally confuses me
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom