8-core SB-E ES Testing & Overclocking

Y

Yamaki

Gawd
Joined
Dec 27, 2002
Messages
741
As some of you know (if you followed the eBay brag thread) I snagged a couple of octocore SB-E Xeons for $300 a pop. After a frightening motherboard accident involving 50 or so bent pins, I finally have the system up and running. I'm working on overcloking it, although I haven't been extremely successful so far. I'm going to need some advice.

Here's the system:
ASUS X79 Sabertooth Mobo
2.3GHz Intel Xeon 8-core QA8E ES (LGA2011)
Noctua NH-U12P SE2 HSF
2x2GB G.Skill 1600MHz DDR3
Antec Earthwatts 380W (all I've got...pretty solid though)
Windows XP Pro (again, all I've got...)
nVIdia 8600GTS OEM (*cough*...yup)

BIOS detected the chip and set the RAM up at 1333MHz, which I'm assuming is in spec. Stock voltage is 1.08V which I think is really good. The max multiplier unfortunately seems to be 24, so I'm stuck with BCLK overclocking.

Results so far:
Stock (24x100, 1.08V):
22C/80W idle
33C/161W load
8.76 Cinebench R11.5

Highest OC (24x108, 1.25V Vcore, 1.2V VCSSA):
32C/132W idle
36C/219W load
9.32 Cinebench R11.5

My current problem is pushing past 2.6GHz (108BCLK). I've pumped up the Vcore and VCCSA voltages to 1.35V and tried every combo I could think of, including ~118MHz BCLK with 125MHz CPU strap, but it really doesn't seem to want to budge from here. Any ideas on how to push it further?

PPD #s coming soon. Thanks for looking! :)

Pics:
Photo%20Dec%2016%2C%206%2038%2048%20PM.jpg

Photo%20Dec%2018%2C%207%2036%2038%20PM.jpg
 
Last edited:
Nope, I went as low as 19x125 (2.4GHz) and it still wouldn't POST. However I'm curious what would happen at 166 strap...
 
Will be keeping an eye on this thread as ordered one of the 4 core SB-EP ES CPU's earlier today. I wonder if these 2QPI CPU's will work in Gigabyte X79 boards.
 
Forget the SB-E, why aren't you folding on that beige beast? :p

I have a copy of Vista x64 I'll give to a fellow folder for free so you don't have to endure Windows XP. Shoot me a PM if you are interested.
 
Well, since I was a bit slow on the draw with regards to getting my post up, I will sum up my findings below. In addition, I vote that we start a new thread about SB-E overclocking performance and motherboard compatibility.

Test setup:

CPUs: 1.6 GHz SB-E QA88 and QA89 ES CPUs
Motherboards: ASUS P9X79 and ASROCK X79 Extreme 4

Overclocking:

I was able to lock the multiplier in at 20x, hitting 2.0 GHz at stock FSB strap. However, I was only able to bump the FSB strap up to ~108 or so, but can't hit any higher despite fiddling with voltages for 45 minutes... :( I'm not sure what we're doing wrong, but it is very saddening if we can't get much higher on these ES CPUs...

With that being said, I am starting to believe that these ES CPUs are of a very early batch, hence the lack of overclocking headroom. Just my 2c.
 
Maybe 125 strap is somehow blocked for Xeon chips? I'm not sure how they would do that though. It would make sense that they wouldn't let 2ghz xeons be run at 3.3.
 
NobleX13 said:
Well, since I was a bit slow on the draw with regards to getting my post up, I will sum up my findings below. In addition, I vote that we start a new thread about SB-E overclocking performance and motherboard compatibility.

Test setup:

CPUs: 1.6 GHz SB-E QA88 and QA89 ES CPUs
Motherboards: ASUS P9X79 and ASROCK X79 Extreme 4

Overclocking:

I was able to lock the multiplier in at 20x, hitting 2.0 GHz at stock FSB strap. However, I was only able to bump the FSB strap up to ~108 or so, but can't hit any higher despite fiddling with voltages for 45 minutes... :( I'm not sure what we're doing wrong, but it is very saddening if we can't get much higher on these ES CPUs...

With that being said, I am starting to believe that these ES CPUs are of a very early batch, hence the lack of overclocking headroom. Just my 2c.
Click to expand...
I can neither confirm or deny that those are from a very early batch...
 
I haven't played with any 2011s yet but if it's anything like 1155 isn't a max BCLK under 110 pretty much the norm for any cpu?
 
Kardonxt said:
I haven't played with any 2011s yet but if it's anything like 1155 isn't a max BCLK under 110 pretty much the norm for any cpu?
Click to expand...

Nope, 2011 is different because it has straps which keep the RAM and PCIE from going out of spec. So the BCLK range is like 100-110 (100 strap), 117-133 (125 strap), an so on. Divide BCLk by strap to get PCIE bus frequency. As long as its between 90-110 it's usually good.

pjkenned said:
Is that "Beige Beast" one of those old school mid 90's Mac clones?
Click to expand...

Yes indeed it is :D the second computer I ever owned (the first was a 386 DOS box). Got it as a gift when I was 6 years old.

Anyway Noble's results confirming mine is pretty disappointing. Looks like these chips will only ever shine in a 2P configuration. I'm not willing to wait out the 2P boards being released nor do I want to spend the money, so I'm probably just going to eBay the chips for profit (unless someone wants them). Such a shame.
 
Zink said:
Does 125 strap work with low multis?
Click to expand...

I've only worked with a couple of X79 motherboards and I was never able to get any of them to POST with any other strap setting than 100MHz.

Kardonxt said:
I haven't played with any 2011s yet but if it's anything like 1155 isn't a max BCLK under 110 pretty much the norm for any cpu?
Click to expand...

Yes. If you are lucky you can hit about 107MHz or so.

Yamaki said:
Nope, 2011 is different because it has straps which keep the RAM and PCIE from going out of spec. So the BCLK range is like 100-110 (100 strap), 117-133 (125 strap), an so on. Divide BCLk by strap to get PCIE bus frequency. As long as its between 90-110 it's usually good.



Yes indeed it is :D the second computer I ever owned (the first was a 386 DOS box). Got it as a gift when I was 6 years old.

Anyway Noble's results confirming mine is pretty disappointing. Looks like these chips will only ever shine in a 2P configuration. I'm not willing to wait out the 2P boards being released nor do I want to spend the money, so I'm probably just going to eBay the chips for profit (unless someone wants them). Such a shame.
Click to expand...

Supposedly you can use a strap setting of 100MHz, 125MHz, 166MHz, or 250MHz. However according to information I've seen passed along by ASUS, you aren't likely to be able to see anything other than a 3-5MHz (7-10MHz according to some sources) increase in BCLK over any strap setting. For all the good it will do as I've yet to be able to get an X79 board to POST using anything other than a 100MHz strap setting.
 
Last edited:
Yamaki said:
Nope, 2011 is different because it has straps which keep the RAM and PCIE from going out of spec. So the BCLK range is like 100-110 (100 strap), 117-133 (125 strap), an so on. Divide BCLk by strap to get PCIE bus frequency. As long as its between 90-110 it's usually good.
Click to expand...

That's all well and good but the PCIE frequency doesn't like to go above 105mhz. This does seem to be the same issue that was seen overclocking the sandy bridge CPUs.
 
osrk said:
That's all well and good but the PCIE frequency doesn't like to go above 105mhz. This does seem to be the same issue that was seen overclocking the sandy bridge CPUs.
Click to expand...

It is. Strap settings don't change that. So far they just give you new ways to prevent your system from POST'ing. It's a feature, not a bug. :D
 
I can see that being true. Although I got Cinebench and LinX to run at 108BCLK, F@H gave me strange results. I was getting 1 hour frame times on a P7903 even though all 16 threads were loaded...

However, in a Tom's Hardware article about X79 they were able to hit 134BCLK at 125 strap with normal i7 chips.
 
I like what our standards are for "OCing well" now. Remember when getting an i7 920 to 4GHz was something to be proud of? On SB/SB-E you're virtually guaranteed 4.5GHz with anything more than stock cooling.
 
EBruddah said:
Outlook not so good on Sandy Bridge - E chips Overclocking...

If they can't get an "X" version stable at 4.8GHz that just means there's no more overhead for SB-E. I'm waiting for IB-E the die shrink and the tri-gate transistors. I think that should OC well. As for now I think we're stuck.

http://hardocp.com/article/2011/12/19/asus_rampage_iv_extreme_lga_2011_motherboard_review
Click to expand...

Yamaki said:
I like what our standards are for "OCing well" now. Remember when getting an i7 920 to 4GHz was something to be proud of? On SB/SB-E you're virtually guaranteed 4.5GHz with anything more than stock cooling.
Click to expand...

So true. I even mentioned that in the Rampage IV Extreme article. Normally I'm used to seeing 5.0GHz+ overclocks now. So it's hard to get excited about 4.6GHz OC's.
 
Yamaki said:
Nope, 2011 is different because it has straps which keep the RAM and PCIE from going out of spec. So the BCLK range is like 100-110 (100 strap), 117-133 (125 strap), an so on. Divide BCLk by strap to get PCIE bus frequency. As long as its between 90-110 it's usually good.



Yes indeed it is :D the second computer I ever owned (the first was a 386 DOS box). Got it as a gift when I was 6 years old.

Anyway Noble's results confirming mine is pretty disappointing. Looks like these chips will only ever shine in a 2P configuration. I'm not willing to wait out the 2P boards being released nor do I want to spend the money, so I'm probably just going to eBay the chips for profit (unless someone wants them). Such a shame.
Click to expand...

I actually just listed mine on eBay; hopefully I didn't influence your decision to embark on this build too much... :( I feel bad.
 
NobleX13 said:
I actually just listed mine on eBay; hopefully I didn't influence your decision to embark on this build too much... :( I feel bad.
Click to expand...

Want to shoot me info via PM?

I am totally bummed... want to test the one I have sitting on my counter...
 
NobleX13 said:
I actually just listed mine on eBay; hopefully I didn't influence your decision to embark on this build too much... :( I feel bad.
Click to expand...

No, not at all, don't feel bad. Despite the disappointing overclocking, it was exciting to play with chips that won't be out for another few months. It's a whole world away from the heavily documented 3930K/etc. overclocking experience. And above all I was shocked at how cool they run despite their 130W TDP--I couldn't even get them to hit 40C!

When all's said and done I'll be making a nice profit over what I originally paid for them :D
 
my 3930k clocks very well to 4.8 at only 1.375v core and 5ghz at 1.425vcore but i leave it at 4.8 as i dont know if 1.425 is to much vcore for the new chips
 
tunatime said:
my 3930k clocks very well to 4.8 at only 1.375v core and 5ghz at 1.425vcore but i leave it at 4.8 as i dont know if 1.425 is to much vcore for the new chips
Click to expand...

1.425v should be OK on these. Though I'm uncertain what kind of temperatures you might end up seeing like that.
 
Last edited:
Bummer about these chips gents. Just wanted to say thanks for being the guinea pigs on this one and saving some of us the aggravation!
 
These Xeon seems to be a very bad trip.

Hope retail chips will accept more overclock .
 
Jeanjean said:
Hope retail chips will accept more overclock .
Click to expand...

I can guarantee they won't since there won't be any. 8c/16t parts are limited to the Xeon series. Best you can get for desktop SB-E parts is 6c/12t.

All we can hope for is that retail Xeons, or maybe even later Q3* or Q4* ES revisions feature unlocked straps like current retail desktop SB-E parts.
 
Dead Things said:
I can guarantee they won't since there won't be any. 8c/16t parts are limited to the Xeon series. Best you can get for desktop SB-E parts is 6c/12t.

All we can hope for is that retail Xeons, or maybe even later Q3* or Q4* ES revisions feature unlocked straps like current retail desktop SB-E parts.
Click to expand...

The Xeon's not taking additional strap settings probably won't make things any worse than they are now. I've never been able to get an X79 based system to boot using anything but a 100MHz base clock. Even 125MHz has been a no go on two out of three boards I've tried it with. I've got another one here but I haven't tried it yet.
 
Dan_D said:
For all the good it will do as I've yet to be able to get an X79 board to POST using anything other than a 100MHz strap setting.
Click to expand...

All I can think of is just make sure changing strap doesn't explode your memory speed. I can interchange 125Mhz and 100Mhz straps without changing anything other than CPU multiplier and Memory speed.
 
Nightbird said:
All I can think of is just make sure changing strap doesn't explode your memory speed. I can interchange 125Mhz and 100Mhz straps without changing anything other than CPU multiplier and Memory speed.
Click to expand...

Yeah I've kept the CPU multiplier and memory speed where they need to be. Neither board I've tried it with would POST like that.
 
lehmann said:
Soo what are the ppd on these chips? just curious.
Click to expand...

Yeah. After the disappointing OC's it seems like people are selling some of them pretty cheap on ebay... still a lot of processor for your money IMO.

Curious what kind of PPD they produce.
 
colinstu said:
Yeah. After the disappointing OC's it seems like people are selling some of them pretty cheap on ebay... still a lot of processor for your money IMO.

Curious what kind of PPD they produce.
Click to expand...

My $0.02 says that these 8 core SB-E CPUs are only worth the money for people needing the kind of threading capabilities they offer. The overclockers may want them, but they'll generally be better served by LGA1155 systems.
 
Yea but you know those 8-core 3.0GHz Xeons in the wild will probably run north of $1k so interesting proposition if you were so inclined.
 
I tried to run PPD numbers but as I said, I was getting 1 hour frame times on P7903. Something was seriously wrong with F@H compatibility. But I didn't have the energy to diagnose it, I just wanted to take the thing apart. Sorry guys. I'd estimate it's about the same as a 2600k @ ~4.8GHz.
 
Yamaki said:
I tried to run PPD numbers but as I said, I was getting 1 hour frame times on P7903. Something was seriously wrong with F@H compatibility. But I didn't have the energy to diagnose it, I just wanted to take the thing apart. Sorry guys. I'd estimate it's about the same as a 2600k @ ~4.8GHz.
Click to expand...

instability and ram come to mind...
 
Dan_D said:
My $0.02 says that these 8 core SB-E CPUs are only worth the money for people needing the kind of threading capabilities they offer. The overclockers may want them, but they'll generally be better served by LGA1155 systems.
Click to expand...

This is the DC subforum, we feast on high thread count! om nom nom. My weaksauce self is only running 16 threads right now, but we have many here with 24, 32, 48, and more recently a few with interlagos 64 threads. If the price on these CPU's is not too high, they could make very nice folding rigs. Even a pair of these at 2.4Ghz would be 32 threads and have some good throughput.
 
You must log in or register to reply here.
Back
Top