Jump to content
HWBOT Community Forums

Recommended Posts

  • Crew
Posted (edited)

This question has been raised several times so I need to clarify this issue. Upon a HWBot member request I've done some research and decided to divide 3700+/4000+ E6 categories from E4 ones concluding the have different cores.
You can grab a coffee and have a in-depth read.
This is my letter to Franck Delattre, the author of CPU-Z who agreed with my arguments and updated CPU-Z to tell them apart.

================
Hello, Franck!
There are two CPUs that aren't documented anywhere and CPU-Z identifies them wrong.
ADA4000DKA5CF and ADA3700DKA5CF, identified as San Diego DH-E6.

Take a look at the table.


image.png.19d108b239aeba2bb41f8da17721c170.png

First, the part number, it's different. Meaning AMD sold it as a different part (they sometimes sold CPUs with different cores under one part number).
Second, the CPUID, it's different but this is because the revision is different.
And now the (al)most interesting - extended model. For E4 it's 27 and for E6 it's 37. But this is not the case for Venice - both E3 and E6 have same 2F model. Meaning it's the same core, but different revisions. And for our 3700/4000+ there are different ext. models.

And the most important - stepping. The stepping code is like a silicon batch.
So here are the stepping codes for several CPUs:

A64 X2 4800+ Toledo: ACBWE CCB2E CCBWE LCB9E LCBIE

A64 4000+ E6: CCBWE LCB9E LCBBE LCBIE

A64 4000+ E4: AABHE CABGE CABHE KAB1E KAB2E KAB3E KABYE KACAE

Summing this up - AMD produced a CPU with a different p/n, different ext. model number and with steppings same a Toledo CPUs. Meaning this indeed is a Toledo core.

Moreover, according to AMD revision guide (where we don't see our CPUs unfortunately):

271289871_.png.73bbe4421ac4497b3cf246062e8f3564.png

but if they were, they'd be in the last line (same core as Toledo) but with their own CPUID (F72) meaning their stepping is JH-E6. JH stands for dual-core die.

I attach you a txt report of such a CPU (single core Toledo).
Please, fix displayed core name and stepping name (JH-E6) for them.
 

Thank you in advance!
Alex
==============================

Here's a photo of such a CPU. You can clearly see that it's much bigger than a regular single-core San Diego. Courtesy of @ultra_code
IMG_20220815_152556816.thumb.jpg.cd25c85b5e373ea6669e991432ed5aa4.jpg

I would like to remind that there are already two other CPUs of such nature:
https://hwbot.org/hardware/processors?key=amd-architecture-k8-athlon_64-athlon_64_3200_(manchester)
https://hwbot.org/hardware/processors?key=amd-architecture-k8-athlon_64-athlon_64_3500_(manchester)

Both aren't present on AMD revision guide.
But if you think it was a simple choice, you're wrong. I've won the battle but definitely didn't win the war. AMD has mixed models and cores in all sorts of ways. We don't have internal info about CPU core names or crystals used so it's almost always an educated guess.
In this case we have several clear signs that divide 3700+/4000+ E4 CPUs from E6.
But here's another one: Venice CA1BE (courtesy of @R-998).

CAB1E.thumb.png.4914491e88b5d494a3f4930e710236c5.png

It has a much bigger core size than a Venice should have concluding it is a San Diego. But there were no E6 San Diegos. Nevertheless, in every way it is identified as a Venice E6. So even if we would've thought about splitting them, we wouldn't be able to detect them correctly.
So not only we need a reason to separate categories one from another, we need a way to detect them correctly which in this case is impossible.

I've done a number of such investigations of obscure cases so let me know if you liked it and would want to read more. Also let me know if you have something valuable to add on this case or other s939 CPUs.
Please, refrain from offtopic.

Edited by Antinomy
  • Like 3
Posted (edited)

Looks like common practice from the factories , back then.
Where at the end of production line , they were using leftover silicon to cover various needs.

Most probably
Toledo cores with defects were used with one core disabled , to match other needs in cpu lines.

 

Edit

Interesting how come we dont have LCBBE in pure Toledo X2 cpu's ... and at the same time we have them as cut-down A64's (3700-4000).

And we also have them in dual core Opterons (180 and 185).

 

 

Edited by TASOS
  • Like 2
  • Crew
Posted
12 hours ago, TASOS said:

Interesting how come we dont have LCBBE in pure Toledo X2 cpu's ... and at the same time we have them as cut-down A64's (3700-4000).

And we also have them in dual core Opterons (180 and 185).

I believe s940 and s939 share the same silicon, just with features disabled (if any). What is interesting is that I've seen an even bigger core than shown and it would be nice to know which dual-cores had bigger cores and which ones had smaller.

Posted

I have LCBBE X2 chips. Seen a few x2 4400+ models with this, and the odd x2 3800+. Never saw an x2 4800+ or FX60 with it, though. Also, Opterons have it, sometimes (less for 939, but plenty of 940 opterons with this type of batch)

I don't really know how part numbers are supposed to work, but there are plenty of other examples - E3 vs E6 venice chips come to mind (they have different part numbers, too, and are distinguishable). Then you have F2 vs F3 Windsor CPUs for AM2, G1 vs G2 Brisbanes (both with different part numbers), and for some Athlon II models you can have Regor, Propus or Deneb cores (Athlon II X2 220 in particular) - but with the same part number. You also have different part numbers for different wattage bins, though I'm not sure if those can be easily identified through software. (like, the 4400+ comes in an ADV and an ADA flavor)

The question is just where you want to draw the line, and I also wonder if this was talked about some 15 years ago, too (it was noticed, and all of us 939 heads knew that these were dual cores with one core disabled - but there was also the question of "category inflation" from what I remember, so the decision was to keep them together)

 

  • Like 3

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...