| View previous topic :: View next topic |
| Author |
Message |
Dome Guest
|
Posted: Sun Apr 12, 2026 10:49 am Post subject: ADA3700AEP5AP |
|
|
Hi,
I’m currently investigating a few AMD Athlon 64 3700+ (ADA3700AEP5AP) processors that show some very unusual but consistent markings.
I understand why these CPUs have been considered suspicious or potentially remarked mobile parts, especially because:
- the data matrix code does not match the visible serial number on top
- some BIOS tools may misidentify them as ES or unknown CPUs
- the model number itself is not clearly documented in some AMD public listings
However, I would like to share a few additional observations from my own sample comparison:
1) The data matrix code appears to be consistent across multiple CPUs of this group, while the top-printed serial numbers differ. This suggests a fixed internal identifier rather than per-unit serialization.
2) There are noticeable differences in laser engraving style (including different fonts for batch/serial information), which are consistent within this specific group but differ from standard Athlon 64 retail CPUs.
3) Physical inspection (heatspreader quality, weight, substrate markings) does not show typical signs of re-marking or aftermarket modification.
At this point, I am still comparing multiple samples and trying to understand whether this represents:
- a special OEM / validation batch,
- a reworked internal AMD production run,
- or something else entirely.
So far, I do not see clear evidence of a typical “remarked mobile CPU” scenario, but I also cannot fully classify them as standard retail parts.
I will continue collecting data and will share updates if I find more consistent patterns or contradictions.
Best regards |
|
| Back to top |
|
 |
KK_Fin

Joined: 01 Jul 2013 Posts: 835 Location: Finland
|
Posted: Sun Apr 19, 2026 4:59 pm Post subject: |
|
|
Interesting, I have tried to investigate this also, but I haven't had really time to do it.
I have two of these "mystery" chips and both are ADA3000AEP5AP. I can confirm your findings. It looks like this is true only for AEP5AP's or at least all mine AEP4AP's, AEP4AR's and AEP5AR's are like common Athlons.
Does anyone here have "normal" AEP5AP? Or are all AEP5AP's like these? _________________ 1242 x86 CPUs in my collection (week 2 2026). |
|
| Back to top |
|
 |
Dome Guest
|
Posted: Wed Apr 22, 2026 11:36 am Post subject: |
|
|
Hi everyone,
I have an update regarding the investigation of the ADA3700AEP5AP (Clawhammer). After extensive testing, I can now confirm that this is indeed a genuine Engineering Sample with unique characteristics that rule out the "remarked mobile CPU" theory.
Key Findings & Stability Data:
Unlocked Multiplier: The most significant discovery is the fully unlocked multiplier, which is adjustable up to 25x. This is a hardware-level feature typical for Engineering Samples and cannot be found on standard retail or mobile chips of this era.
Extreme Undervolting/Efficiency:
800 MHz @ 0.8V: The CPU runs rock-stable at only 0.8V. 2200 MHz @ 1.35V: It completes long-term Prime95 sessions at 2.2 GHz with just 1.35V (well below the standard 1.5V for retail Clawhammers), staying remarkably cool at 39-41°C.
Overclocking Potential: Initial tests showed stability at 2.4 GHz @ 1.5V, and even 2.6 GHz was reachable for short periods under air cooling, only limited by the motherboard's voltage stability.
Best regards |
|
| Back to top |
|
 |
xsecret

Joined: 01 Feb 2004 Posts: 1846 Location: France
|
|
| Back to top |
|
 |
Guest
|
Posted: Thu Apr 23, 2026 3:03 am Post subject: |
|
|
To the user asking for CPU-Z: I tried to register to upload the images or post the link directly, but the bot detection is blocking my IP.
The multiplier is unlocked up to 25x. I tested 2.6GHz.
Name: AMD Athlon 64
Code Name Clawhammer
Package Socket 754
Technology 0.13
Core Vorlage 1.2 V
Specification AMD Engineering Sample (ES)
Family F Model 4 Stepping 8
Ext. Family F Ext. Model 4 Revision SH7-C0
Instructions MMX(+), 3DNow!(+), SSE, SSE2, x86-64
Core Speed 800.96
Multiplier 4
Bus Speed 200.24MHz
L1 Data 64 KBytes 2-way
L1 Inst. 64 KBytes 2-way
Level 2 1024 KBytes 16-way |
|
| Back to top |
|
 |
Wasmachineman_NL

Joined: 04 Jul 2019 Posts: 988 Location: Netherlands
|
Posted: Thu Apr 23, 2026 5:50 am Post subject: |
|
|
>To the user asking for CPU-Z: I tried to register to upload the images or post the link directly, but the bot detection is blocking my IP.
Peak irony as I suspect OP of using an LLM to write his posts. |
|
| Back to top |
|
 |
Dome Guest
|
Posted: Thu Apr 23, 2026 6:11 am Post subject: |
|
|
| Yes, I am using LLM because my English ist very bad. So? |
|
| Back to top |
|
 |
xsecret

Joined: 01 Feb 2004 Posts: 1846 Location: France
|
|
| Back to top |
|
 |
Dome Guest
|
Posted: Thu Apr 23, 2026 12:46 pm Post subject: |
|
|
| The one from Reddit is mine. That ist the one with unlocked multiplier. Runs stable 800MHz @ 0.8V, 2,2GHz @ 1.35V and 2.4GHz @ 1.55V. |
|
| Back to top |
|
 |
Dome Guest
|
Posted: Thu Apr 23, 2026 1:00 pm Post subject: |
|
|
I'm currently blocked from registering, so I'll provide the proof this way. Here is the visual confirmation of the Clawhammer ES status.
Link: ibb [DOT] co [SLASH] Xf81vwhZ |
|
| Back to top |
|
 |
xsecret

Joined: 01 Feb 2004 Posts: 1846 Location: France
|
Posted: Fri Apr 24, 2026 8:37 am Post subject: |
|
|
Use a newer CPU-Z version (it works on XP) and click on "Validate", then report the URL here.
PS : There is no "ES" flag on AMD CPUs (vs Intel CPUs) and the PNS string is written by BIOS. It's very common to write "AMD Engineering Sample" when the BIOS is not able to recognize a newer CPU (even a retail one). _________________ ES-Only Collector : http://www.engineering-sample.com
Universal Chip Analyzer (UCA) : https://x86.fr/uca / http://www.cpu-world.com/forum/viewtopic.php?t=34349 |
|
| Back to top |
|
 |
Dome Guest
|
Posted: Fri Apr 24, 2026 9:12 am Post subject: |
|
|
While the BIOS string can be influenced by the microcode, CPU-Z identifies the "Engineering Sample" status and the specific "SH7-C0" revision by reading the CPUID registers directly. This is not a generic BIOS string for an unknown CPU.
A CPU-Z validation will be provided in due time (my mainboard just died and I am waiting for another one). However, a standard retail Athlon 64 is hard-locked, while this specific sample allows a multiplier of up to 25x — which is the definitive proof that this is not a remarked retail chip. |
|
| Back to top |
|
 |
Dome Guest
|
Posted: Fri Apr 24, 2026 9:25 am Post subject: |
|
|
| I agree that CPU-Z alone is not a definitive proof of engineering sample status, and I will provide a full validation link once my system is operational. My current conclusion is based on observed multiplier behavior and voltage scaling characteristics, which are not consistent with standard locked retail Clawhammer parts, but I acknowledge that final classification requires full CPUID validation. |
|
| Back to top |
|
 |
Dome Guest
|
Posted: Sat Apr 25, 2026 1:13 pm Post subject: |
|
|
| h t t p s : / / valid.x86.fr/5jfxmu |
|
| Back to top |
|
 |
socket
Joined: 25 Jun 2011 Posts: 16
|
Posted: Sun Apr 26, 2026 4:51 pm Post subject: |
|
|
What exactly does this validation prove? Only thing we know for a long time - this fake CPU is (in majority of motherboards) running as 800 MHz sample with x4 multiplier, and in some other boards it's not even posting.
I personally have one piece btw. |
|
| Back to top |
|
 |
|