Xeon Owners Club

[phill]

I thought I had a bit of a better reason to post in here again, I found a few Xeon's laying about

Here's a few more pics

64Gbs of lovely ECC DDR3

I upgraded the bios so it would take the V2 CPUs....

Just testing with some Windows 10 at the moment, making sure it all works When I get the other heatsink through from my good friend @TheMadDutchDude I'll be seeing how 20C and 40T gets along

And sadly as I didn't read the model numbers correctly I had two 8C 8T CPUs... I'm hoping they can go back but

I couldn't resist Oh and here's just some more lunacy...

What's better than 2 12C 24T CPUs??

You guessed it.....

4 12C 24T CPUs
Two of these will make it over to me in the UK but two might get sold on.. My good friend might have to sell on but we'll see

I hope the pics are good enough for the thread

 

[natr0n]

Ever since my backup nic got fried and usb wifi is meh,so been using the xeon system 24/7. I cant go back to the 2600k. Horsepower is awesome.

 

[Darmok N Jalad]

I never owned a slocket system. I didn't even know there were Xeons that ran on that platform, but I guess it makes sense since it was all about the cache. My college roommate had a PII 233MHz Gateway2000 though. It was pretty crazy to open the case and see that big chip and heatsink on them. So those are 500MHz Xeons?

 

[lexluthermiester]

Definitely going to hold on to them and keep an eye out for a dual or quad slot 2 motherboard. That'll be a long-term side project though, obviously they're super rare.

I've never actually seen a quad Slot2 board before. Duals popup on ebay from time to time, example;

Dell 00024048 Precision motherboard for Dual Slot 2 Xeon processors. 5PCI, 1ISA, for sale online

Find many great new & used options and get the best deals for Dell 00024048 Precision motherboard for Dual Slot 2 Xeon processors. 5PCI, 1ISA, at the best online prices at eBay! Free shipping for many products!

www.ebay.com

 

[Easy Rhino]

How are those scalable chips? Been looking at the scalable silver for a new home lab.

 

[Edwired]

It been pretty quiet here did somebody's xeon system blow up yet?

 

[Easy Rhino]

It been pretty quiet here did somebody's xeon system blow up yet?

Well I asked about those scalable chips but nobody seemed to care

 

[Edwired]

Well I asked about those scalable chips but nobody seemed to care

I know the feeling sometime people dont have the answer to the question

 

[phill]

Well I asked about those scalable chips but nobody seemed to care

I'm going to be really daft at this point and ask, what are scalable chips??

 

[MrGRiMv25]

I think it's a load of extra hardware functions for AI, data center and HPC uses, AVX has a couple extra functions over standard Xeons too - there's a breakdown in a couple of the links below.

Intel Xeon Scalable - What you need to know - Thinkmate

Our guide on Intel's biggest data center launch in a decade.

www.thinkmate.com

The Intel Second Generation Xeon Scalable: Cascade Lake, Now with Up To 56-Cores and Optane!

www.anandtech.com

Quick list without opening links:

• Most of the mid-range processors have more cores for the same price
• Frequency has increased in almost all processors
• L3 has increased in most mid-range processors
• Faster DDR4 is supported
• More DRAM is supported across the stack
• Optane DRAM is supported on almost all Gold/Platinum SKUs
• New CPU configurations optimized for specific markets
• Speed Select Technology for Cloud Deployments
• New ‘Cascade Lake-AP’ Platinum 9200 family, up to 56 cores and 400W per socket
• Enhanced Spectre and Meltdown Mitigations
• New AVX-512 VNNI instructions for Emerging Workloads

 

[phill]

I think it's a load of extra hardware functions for AI, data center and HPC uses, AVX has a couple extra functions over standard Xeons too - there's a breakdown in a couple of the links below.

Intel Xeon Scalable - What you need to know - Thinkmate

Our guide on Intel's biggest data center launch in a decade.

www.thinkmate.com

The Intel Second Generation Xeon Scalable: Cascade Lake, Now with Up To 56-Cores and Optane!

www.anandtech.com

Quick list without opening links:

• Most of the mid-range processors have more cores for the same price
• Frequency has increased in almost all processors
• L3 has increased in most mid-range processors
• Faster DDR4 is supported
• More DRAM is supported across the stack
• Optane DRAM is supported on almost all Gold/Platinum SKUs
• New CPU configurations optimized for specific markets
• Speed Select Technology for Cloud Deployments
• New ‘Cascade Lake-AP’ Platinum 9200 family, up to 56 cores and 400W per socket
• Enhanced Spectre and Meltdown Mitigations
• New AVX-512 VNNI instructions for Emerging Workloads

Ah, I get you The Xeon equivalent of X299

 

[8BitZ80]

Anyone else had experiences with the Xeon L5420? Mine isn't overclocked and runs at ~65C idle and touches 90C during a Cinebench run. I tried different coolers to make sure it wasn't a thermal contact issue, but nope, it just seems to be running terrible despite the 50W TDP. Not sure what the deal is.

The X3363 for comparison has an 80W TDP and idles in the high 40s with a max of almost 80. Performs better, too.

 

[phill]

I've got a few of the L5640's and they run very cool.. Even when running 100% loaded with World Grid Computing they struggled to hit 40C in my EVGA SR-2 system..

When it's actually saying those temps, can you feel a lot of heat coming off the heatsink or anything like that? What coolers are you using with the CPU?

 

[8BitZ80]

I've got a few of the L5640's and they run very cool.. Even when running 100% loaded with World Grid Computing they struggled to hit 40C in my EVGA SR-2 system..

When it's actually saying those temps, can you feel a lot of heat coming off the heatsink or anything like that? What coolers are you using with the CPU?

Stock Intel heatsinks, they do the job. All my thermal numbers are while using those coolers.
It's possible that it's just bad seating, but I'll need to set up another test bench to verify.

 

[phill]

I was just wondering if the sensor on the CPU was acting up? I don't think it would be unheard of. If my little Googling is correct, it's a 771 socket, so it could be something in the socket?? Have you tested another CPU in there or anything like that?

 

[8BitZ80]

I was just wondering if the sensor on the CPU was acting up? I don't think it would be unheard of. If my little Googling is correct, it's a 771 socket, so it could be something in the socket?? Have you tested another CPU in there or anything like that?

Believe it or not, I think this CPU has an issue with the internal IHS solder. I tested all the different configurations, even holding a tower cooler down by hand to check the contact, but all the core temperatures read consistently high and the cooler doesn't get hot. No question now that it's faulty.
I may try breaking it open with the vice to see what I find, for science. It only cost 6 USD so it's no big loss.

 

[phill]

Sounds like a definite option and way forward as well

Please let us know your findings

 

[Dinnercore]

Believe it or not, I think this CPU has an issue with the internal IHS solder. I tested all the different configurations, even holding a tower cooler down by hand to check the contact, but all the core temperatures read consistently high and the cooler doesn't get hot. No question now that it's faulty.
I may try breaking it open with the vice to see what I find, for science. It only cost 6 USD so it's no big loss.

There is a chance that these 771 chips are not even soldered. From my findings, in such a case the TIM underneath the IHS is very brittle and dry due to the age of the CPU. It could be that this paste cracked, due to thermal stress or from dropping it. Maybe someone even did a delid and resealed it but did a poor job on the paste.
Either way, I´d be very interested to see what´s underneath, please let us know

 

[8BitZ80]

There is a chance that these 771 chips are not even soldered. From my findings, in such a case the TIM underneath the IHS is very brittle and dry due to the age of the CPU. It could be that this paste cracked, due to thermal stress or from dropping it. Maybe someone even did a delid and resealed it but did a poor job on the paste.
Either way, I´d be very interested to see what´s underneath, please let us know

Looks like it wasn't soldered, so I've learned something new today. It came apart pretty easy. The TIM looks shiny and it's super tacky, not sure what it is though.
For giggles I cleaned up all the parts with a razor blade and put it together with some cheap GD900 (not a great internal thermal paste, but anyway).
Hardly made a difference though, the new thermals are pretty much the same.

EDIT: Picture...

Spoiler: Delidded Xeon

 

[Dinnercore]

Looks like it wasn't soldered, so I've learned something new today. It came apart pretty easy. The TIM looks shiny and it's super tacky, not sure what it is though.
For giggles I cleaned up all the parts with a razor blade and put it together with some cheap GD900 (not a great internal thermal paste, but anyway).
Hardly made a difference though, the new thermals are pretty much the same.

EDIT: Picture...

Spoiler: Delidded Xeon

View attachment 135204

Oh that is definitly a soldered chip! Sorry to hear that thermals did not improve much.

Did you completly remove the indium solder from the die and IHS? It´s best to have very shiny surfaces before applying paste and putting the lid back on. Did you remove the black glue around the sides too? This glue can cause height problems where the IHS does not make good contact with the die.

Alternativly you can try to carefully run direct die if you know the cooler and mainboard fit without shorting anything or hitting caps around the socket. Just be very careful if you try, you will need to remove the lever-mechanism of the socket. This way you would know if maybe the sensors are reporting wrong temps or if the trouble comes from something else

EDIT: I have a thread about delidding soldered chips, if you want to see how I treated mine: CLICK ME

 

[8BitZ80]

Oh that is definitly a soldered chip! Sorry to hear that thermals did not improve much.

Did you completly remove the indium solder from the die and IHS? It´s best to have very shiny surfaces before applying paste and putting the lid back on. Did you remove the black glue around the sides too? This glue can cause height problems where the IHS does not make good contact with the die.

Alternativly you can try to carefully run direct die if you know the cooler and mainboard fit without shorting anything or hitting caps around the socket. Just be very careful if you try, you will need to remove the lever-mechanism of the socket. This way you would know if maybe the sensors are reporting wrong temps or if the trouble comes from something else

EDIT: I have a thread about delidding soldered chips, if you want to see how I treated mine: CLICK ME

That's solder? No way! I assumed that solder will rip the silicon apart if you try to delid it. Well there you go, now I'm doubting if that holds any truth at all.
And yeah I cleaned off all the IHS glue, I've done delidding before. Never dared to touch a soldered chip though so this is new to me.

 

[Dinnercore]

That's solder? No way! I assumed that solder will rip the silicon apart if you try to delid it. Well there you go, now I'm doubting if that holds any truth at all.
And yeah I cleaned off all the IHS glue, I've done delidding before. Never dared to touch a soldered chip though so this is new to me.

Yes that is indium solder, very soft metal. Because it is so soft it does not rip the silicon apart if you do it right. Like you did with the vice, you can just apply pressure at a very flat angle and push it off. As long as you keep the angle flat and don´t use sudden force (like using a screwdriver and a hammer to pry it off) you can delid many soldered chips without using heat. I´ve done this 4x with a vice so far and all CPU´s survived.

 

[Edwired]

Anyone else had experiences with the Xeon L5420? Mine isn't overclocked and runs at ~65C idle and touches 90C during a Cinebench run. I tried different coolers to make sure it wasn't a thermal contact issue, but nope, it just seems to be running terrible despite the 50W TDP. Not sure what the deal is.

The X3363 for comparison has an 80W TDP and idles in the high 40s with a max of almost 80. Performs better, too.

Dont forget there was the tjmax offset in those 771 xeon listed here

45nm Xeon Quad-Core CPUs Model Tj Max (Stepping unspecified)
E7440, E7430, E7420 90°C
L7445 80°C
X54xx series 85°C
E54xx series 85°C
L5408 95°C
L5430, L5420, L5410 70°C
X33xx 95°C
L3360 90°C
X33xx series 95°C
L3360 90°C

Download hwinfo64 and see if the difference shown between hwmonitor and hwinfo64. as the tjmax on the xeon e5450 i have is 85c at the moment it fairly cold here it 7c outside and the coldest core reading was 7c on core 2 for some reason core 1 seems to have a stuck sensor as it stayed at 23c all day overrall i had max temp 38c the other day. but if i add more vcore the core temp goes up as normal with the rest of the temp sensors

Yes that is indium solder, very soft metal. Because it is so soft it does not rip the silicon apart if you do it right. Like you did with the vice, you can just apply pressure at a very flat angle and push it off. As long as you keep the angle flat and don´t use sudden force (like using a screwdriver and a hammer to pry it off) you can delid many soldered chips without using heat. I´ve done this 4x with a vice so far and all CPU´s survived.

I say abit of butane torch and vice would work as you soften the solder and the glue the same time but dont torch it for long as i did that once time and the die came away with the ihs.

Bonus tip is to lap the ihs until it flat with the heatsink as some have the inward bow or outward bow in the ihs as some would cause weird reading in the temperature

 

[8BitZ80]

Dont forget there was the tjmax offset in those 771 xeon listed here

45nm Xeon Quad-Core CPUs Model Tj Max (Stepping unspecified)
E7440, E7430, E7420 90°C
L7445 80°C
X54xx series 85°C
E54xx series 85°C
L5408 95°C
L5430, L5420, L5410 70°C
X33xx 95°C
L3360 90°C
X33xx series 95°C
L3360 90°C

Download hwinfo64 and see if the difference shown between hwmonitor and hwinfo64. as the tjmax on the xeon e5450 i have is 85c at the moment it fairly cold here it 7c outside and the coldest core reading was 7c on core 2 for some reason core 1 seems to have a stuck sensor as it stayed at 23c all day overrall i had max temp 38c the other day. but if i add more vcore the core temp goes up as normal with the rest of the temp sensors


I say abit of butane torch and vice would work as you soften the solder and the glue the same time but dont torch it for long as i did that once time and the die came away with the ihs.

Bonus tip is to lap the ihs until it flat with the heatsink as some have the inward bow or outward bow in the ihs as some would cause weird reading in the temperature

I noticed these Xeons can sometimes have temperature reading issues. Some cores can read lower than others. My suspicion was that the sensors have just gone out of spec from long term use, either that or the outer cores in the chip layout don't get good heat dissipation. Depends on how the physical cores are numbered.

 

[Grog6]

If there's solder on the chip, and you put HS compound on top, you'll get terrible temperatures.

The indium solder is so jagged on a microscopic scale, that it doesn't make good contact.

Scraping it off with a razor blade until you get to the die, and then lapping it with 1000 or 2000 grit wet-or-dry emery cloth will make a nice difference.

I didn't realize there was a Xeon version of the Pentium D chips.