I have a pair of Intel E5340 Xeon ES processors I recently acquired and installed in a system the other day. The performance is much less than lower frequency retail Xeon processors I have used in my other systems. Is there a major performance difference between Intel engineering sample processors and their release products? I find it strange because if so, why would hardware review sites be given ES products when they wouldn't accurately represent the actual performance of the processor? Could there be something else I overlooked? I expected much better performance than I'm receiving, at least 25% better if not more.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
-
Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.
Intel Engineering Sample processors and performance
- Thread starter APOLLO
- Start date
They should perform the same. I've used half a dozen or more ES processors from Intel and the performance on them was identical to that of retail processors.
Thanks Dan for the very quick reply. I guess it must be something about the board then. Perhaps the BIOS settings? I'm shocked how low the performance is, very strange.
Must be. Fortunately, I have a spare from another manufacturer that I can swap if it turns out to be the board. I should know in a day or two after my testing is through.
Know someone who can get them.
Or just use ebay.
some of the stories about ES chips..some rumours, some true:
1) ES chips are supplied to reviewers and are the pick of the litter- sounds logical and possibly true.
2) ES chips are tested hard in the intel labs- again, sounds logical and possible true and might explain why someone getting a second hand chip ES chip has a tough time getting it to perform well..just slowly on the way to worn out..you see some ES chips that look like they've been through the ringer on ebay.
3) ES chips are usually early stepping and don't overclock as well as later releases- I believe that to be very true. i imagine there is ES versions at every stepping. it makes sense that the further down the evolution of a processor, especially a new model, that it should be "better".
I have an ES B1 stepped 9450. it's good, will do 3400 (8 x 425) or 7.5 x 465ish stable on my abit ip35 pro but it's not going to ever get close to 4000 that i can envision. how much of this is the immature bios on my abit board? how much of it is the processor stepping..how much of it is simply that the quad core wouldn't be pushed to that high an fsb? i anxiously await people running a C0 or M1 stepping 9450 on my board version to see how they do
1) ES chips are supplied to reviewers and are the pick of the litter- sounds logical and possibly true.
2) ES chips are tested hard in the intel labs- again, sounds logical and possible true and might explain why someone getting a second hand chip ES chip has a tough time getting it to perform well..just slowly on the way to worn out..you see some ES chips that look like they've been through the ringer on ebay.
3) ES chips are usually early stepping and don't overclock as well as later releases- I believe that to be very true. i imagine there is ES versions at every stepping. it makes sense that the further down the evolution of a processor, especially a new model, that it should be "better".
I have an ES B1 stepped 9450. it's good, will do 3400 (8 x 425) or 7.5 x 465ish stable on my abit ip35 pro but it's not going to ever get close to 4000 that i can envision. how much of this is the immature bios on my abit board? how much of it is the processor stepping..how much of it is simply that the quad core wouldn't be pushed to that high an fsb? i anxiously await people running a C0 or M1 stepping 9450 on my board version to see how they do
My problem is not OC, they are quite decent in that regard. I don't know how much Xeons could have been abused since few server boards support OC and the CPU multiplier seems to be adjustable only downwards.
The issue I have is with actual performance. It's freaking weird. I increase the frequency and performance decreases not increases. It's almost as if some kind of throttling is occurring but I know thermal management is turned off in the BIOS. I have never experienced this kind of anomaly before in all my years of system building.
Clock for clock, these ES processors are performing far, far less than retail Xeons I have in another system. Same frequency, yet the performance is about 30% less. Someone explain this phenomenon to me.
The issue I have is with actual performance. It's freaking weird. I increase the frequency and performance decreases not increases. It's almost as if some kind of throttling is occurring but I know thermal management is turned off in the BIOS. I have never experienced this kind of anomaly before in all my years of system building.
Clock for clock, these ES processors are performing far, far less than retail Xeons I have in another system. Same frequency, yet the performance is about 30% less. Someone explain this phenomenon to me.
that one i don't know about, apollo aside from early revision with errata is my guess causing the performance issues. explain to me why my sensors are all stuck on my es9450 and it shows either 221C on all cores at all times or -30C on all cores..loaded or unloaded..
oh..i know why..early stepping and i'm the guinea pig
ray...i think some used to have unlocked multis..haven't seen any of the 45nm penryn ES's with such a thing though.
oh..i know why..early stepping and i'm the guinea pig
ray...i think some used to have unlocked multis..haven't seen any of the 45nm penryn ES's with such a thing though.
What board are you running them in? How are you overclocking them, pad mod or in the bios? What are your individual core temps, use HWmonitor from CPUID.com.
The board I'm currently running the processors in is a Supermicro X7DAL-E. I am employing three simultaneous methods for OC. BSEL mod on the pads to increase the default FSB from 1066MHz 1333MHZ. Multiplier is then reduced from 9x to 8x in the BIOS because 3GHz didn't seem stable with the stock HSFs. Lastly, I used SysTool to increase the FSB incrementally in Windows.
HWmonitor reports core temps per the following:
Temperature sensor 0 26°C (78°F) [0x4A] (core #0)
Temperature sensor 1 25°C (76°F) [0x4B] (core #1)
Temperature sensor 2 22°C (71°F) [0x4E] (core #2)
Temperature sensor 3 20°C (67°F) [0x50] (core #3)
Temperature sensor 0 15°C (58°F) [0x55] (core #0)
Temperature sensor 1 22°C (71°F) [0x4E] (core #1)
Temperature sensor 2 23°C (73°F) [0x4D] (core #2)
Temperature sensor 3 23°C (73°F) [0x4D] (core #3)
An interesting thing I noticed from HWmonitor is the fact it designates the processors as X5340. The seller referred to them as E5340.
The temps are strange and the processors are at full load working on two SMP clients at the moment. I am going to shut things down and see what the board identifies the processors as. I don't know if the BIOS is the latest. I purchased the board used, but will check on SM's site.
My temps on the cores range between 63-68C when running dual SMP clients on my X5355s. I'm wondering if you're getting throttling from either your processors or FB-Dimms because they're running too hot. Hows it perform at stock?
Some throttling must be occurring despite my apparently futile attempt to prevent that. The reason I believe this is happening stems from the fact my frame times vary even though the system is doing nothing else. None of my other systems do that.
The BIOS doesn't report the actual CPU model, but CPU-Z and CrystalCPUID both identify the processors as X5340. I don't know the difference between that and an E5340. I think the only thing I have left to do now is go through the laborious task of swapping boards and try the processors on a Tyan Tempest. If I don't get any anomalies then it's definitely an issue with the board
The TDP is higher on the X series then the E series, I think it's 120W vs 80W, so generally, those chips will run hotter. Keep an eye on your FB-Dimm temps as well.
I'd also try them at stock, overclocking them probably isn't going to work out to well unless you get some great cooling.
I'd also try them at stock, overclocking them probably isn't going to work out to well unless you get some great cooling.
If i did, I don't remember seeing it, I'll check the next time I reboot.
Let me know if you decide to get rid of one of those boards, I need something for home server and have a pair of 5060s that need a home.
I don't think all ES are the same.
The ones that go into the hands of reputable review sites such as Anandtech and [H] are cherry picked. The review results matches with few people who got lucky purchasing the retail version that matches the review result; requires low voltage for high overclocking.
p.s. oops, I guess Anandtech got luckier than [H]. ;-)
The ones that go into the hands of reputable review sites such as Anandtech and [H] are cherry picked. The review results matches with few people who got lucky purchasing the retail version that matches the review result; requires low voltage for high overclocking.
p.s. oops, I guess Anandtech got luckier than [H]. ;-)
You may be correct about all ES processors not being created equal, but that doesn't explain why the performance decreases when I increase the processors' clock speed substantially. In any case, they're already in the other board awaiting testing. Will post my findings later today.
Come to think of it, I distinctly recall some time ago about another poster who had a similar story concerning his processor. I don't know if it was in the Intel forum. He also claimed that raising the frequency didn't improve performance. I wonder if we have similar issues?