• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD THREADRIPPER VS INTEL SKYLAKE X

sidthesexist said:
Will be even more interesting if there's a way to unlock those non used ones as well?
Click to expand...

Absolutely. I'm still trying to get my head round any other possible reason for AMD to "give away" 2 cores for free. On the face of it, it makes no sense unless they plan on selling 32 core's at a latter stage with an option for current owners to "upgrade" to 32 core's via a paid for bios update ?
 
kitfit1 said:
Absolutely. I'm still trying to get my head round any other possible reason for AMD to "give away" 2 cores for free. On the face of it, it makes no sense unless they plan on selling 32 core's at a latter stage with an option for current owners to "upgrade" to 32 core's via a paid for bios update ?
Click to expand...


Could be just epyc chips that have a failure on one die? or just cheaper to manufacture on one production line similar to the early 4** series graphics cards that were sold as 4 gig but actually had 8?
 
Probably epyc chips with one or two faulty dies. They can likely test the dies individually before building the chip but there may be some infinity fabric fault that they can't detect until they build the whole chip.
 
I'll be surprised if any cores are unlockable - with the architecture they can just bin packages for appropriate market segments - the number of fully assembled CPUs that have partial failures should be very small and any binning process fairly granular so less need to leave room for tolerances.
 
DarkHorizon472 said:
what settings are you using ?

So far no one has posted how to get corsair lpx 3200 single side Hynix to run at 3200, so that would be very helpful.
Click to expand...

Gigabyte k7, F4 bios

3.9OC @1.4v

Selected profile 1 within XMP menu, rebooted .

I think the timings default are
15
17
17
17
36

Did the same thing of a r6 1600 build with LPX 3200 on a 350 gaming(gigabyte) month or so ago

Forget the timings on that.
 
humbug said:
Murderer! :O
Click to expand...

Totally agree. But on the other side intrigued with the 4 cores under there, which means is just an up clocked EPYC 7351 (as parts going) without the 128 pci-e lanes for usage, and the ability to use the secondary 128 pci-e for dual CPU setup.

That would probably means that even at the price the 32 core CPU is sold (far less than the Intel Xeons), AMD is making a humongous profit on them (and rightfully so).

Kudos to them imho.


And as someone put it on the der8auer video

Intel: Look we have 18 cores, much better than 16 cores.
AMD: That is cute, here we have a 32 core CPU. :P
Intel: Back to bribing OEM's I guess...
Click to expand...
 
Comment pointing at the top left capacitor on the board when he showed a still, it's been knocked off its pads.

Next time round get a long thin razor blade to push down the middle to release the central glue strip and it will need less brute force :)
 
Hotwired said:
Comment pointing at the top left capacitor on the board when he showed a still, it's been knocked off its pads.
Click to expand...

Those are normally just for power filtering (decoupling to ground) and shouldn't stop it working just give them better tolerance to a wider range of conditions.

EDIT: Dunno how long they were heating it for but if you are getting it warm enough to loosen up solder you probably don't have more than about 10 seconds at full heat before those dies start taking damage.
 
Last edited:
Rroff said:
EDIT: Dunno how long they were heating it for but if you are getting it warm enough to loosen up solder you probably don't have more than about 10 seconds at full heat before those dies start taking damage.
Click to expand...

Being as young as he is i'm sort of thinking he has little to no experience of de-lidding a soldered lid. It needs to be done very fast and with the maximum heat possible at the start. Going at it over 3 sessions in a gingerly mode is almost certainly what killed it.
Having de-lidded and killed quite a few in the past, at my own expense, it's something you learn to do do very fast lol.
 
kitfit1 said:
Being as young as he is i'm sort of thinking he has little to no experience of de-lidding a soldered lid. It needs to be done very fast and with the maximum heat possible at the start. Going at it over 3 sessions in a gingerly mode is almost certainly what killed it.
Having de-lidded and killed quite a few in the past, at my own expense, it's something you learn to do do very fast lol.
Click to expand...

My experience comes more from SSOP-28 soldering :s but some overlap - likewise generally best to use a high heat and be quick and precise - killed a few chips from having to redo them being too cautious in approach and conservative with heat.
 
The process was fine, the problem was that no one knew that amd was going belt and bracers by having a central glued strip as well as around the edges.

Heating the solder was not working because the central strip was still holding it down securely.
 
Hotwired said:
The process was fine, the problem was that no one knew that amd was going belt and bracers by having a central glued strip as well as around the edges.

Heating the solder was not working because the central strip was still holding it down securely.
Click to expand...
Hotwired said:
The process was fine, the problem was that no one knew that amd was going belt and bracers by having a central glued strip as well as around the edges.

Heating the solder was not working because the central strip was still holding it down securely.
Click to expand...

Don't be silly. Opteron's had the same thing in between dies as well. It always came down to how much heat you used...........or didn't.
None of this is new, what is new though is that none of us knew that threadripper had 4 cores, so the sticky in the middle is slightly longer than it was with Opteron's.
 
Panos said:
Totally agree. But on the other side intrigued with the 4 cores under there, which means is just an up clocked EPYC 7351 (as parts going) without the 128 pci-e lanes for usage, and the ability to use the secondary 128 pci-e for dual CPU setup.

That would probably means that even at the price the 32 core CPU is sold (far less than the Intel Xeons), AMD is making a humongous profit on them (and rightfully so).

Kudos to them imho.


And as someone put it on the der8auer video
Click to expand...


Intel: Look we have 18 cores, much better than 16 cores.
AMD: That is cute, here we have a 32 core CPU. :p
Intel: Back to bribing OEM's I guess...
Click to expand...

:D
 
You must log in or register to reply here.
Back
Top Bottom