I'm gonna guess 5D4 .... what do I win?

By the way stumbled on this, every K6 class datasheet you might want...
https://www.ardent-tool.com/CPU/Docs_AMD.html#K6

It says 5D0. I don't know is it just me but CPU-Z flips around on cpu init.

I can now confirm 630MHz (6x105) 2.1V is stable. Temperature is around 40-50°C with Intel (Mendocino?) cooler. Lid is still on my table.

RaiderOfLostVoodoo wrote on 2022-02-22, 12:17:

Step 1: Order dozens of K6-2+/570ACZ for 20€ each Step 2: Order fake lids that read K6-III+/550ACR Step 3: ? Step 4: Profit! […]
Show full quote

Step 1: Order dozens of K6-2+/570ACZ for 20€ each
Step 2: Order fake lids that read K6-III+/550ACR
Step 3: ?
Step 4: Profit!

I actually kinda had this idea, to make for lyself a unofficial k6-III+ 600EE/FAK/OVC whatever using a early k6 lid that used ink to print the logo and a laser etcher. But the early lids were slightly different than later lids and so far we haven’t conclusively discovered the exact adhesive used. I would want it to look as legit as possible for my collection :p

fool wrote on 2022-02-22, 17:26:

I can now confirm 630MHz (6x105) 2.1V is stable. Temperature is around 40-50°C with Intel (Mendocino?) cooler. Lid is still on my table.

Yep it seems we have successfully brought k6 3+ 550 level performance/overclocks to the masses. The 570 chips are definitely the ones to do this to.

So has anyone tested what happens if instead of moving the 0 ohm resitor it's just taken away? Will it report 0 kb of cache?

Am386DX-40 wrote on 2022-02-22, 19:45:

So has anyone tested what happens if instead of moving the 0 ohm resitor it's just taken away? Will it report 0 kb of cache?

Legends say that if you connect all resistors you’ll end up with a magical AMD K6-IV with 384KB L2-cache 😜.

Serious note:
Would be interesting to see if the chip will overclock higher with 0KB L2-cache 😀.

Thandor wrote on 2022-02-22, 21:14:

Would be interesting to see if the chip will overclock higher with 0KB L2-cache 😀.

There wouldn't be much point in that. Lack of L2 cache will more than offset any gains from overclocking. Think Celeron-266 vs. P2-266.

stamasd wrote on 2022-02-22, 21:21:

Thandor wrote on 2022-02-22, 21:14:

Would be interesting to see if the chip will overclock higher with 0KB L2-cache 😀.

There wouldn't be much point in that. Lack of L2 cache will more than offset any gains from overclocking. Think Celeron-266 vs. P2-266.

Of course but I wonder if overclocking that last 50MHz is fighting against L2-cache or still the architectural limits of the K6.

On quite a few super 7 boards, you're fighting the limits of the L2/L3 to push the bus speed. So without onboard cache that's a bit special, cutting the on die cache might have you not getting very far. Some folks have managed higher clock without the motherboard cache, but that also contributes to overall performance by a few percent, so the few percent ahead on one thing doesn't make up much for the few percent behind on another..

Here's the spreadsheet of a guy who has done a lot of top end tinkering with cache sizes and core types so you can extrapolate yourself what which gets you...
https://drive.google.com/file/d/12_JgRPXbcj8j … 8-i7KYLC3w/view

Doesn't matter if it's actually faster to use, it would just be *interesting* to see if the core can clock higher without the L2. Too bad it doesn't support any higher multiplier than 6x, so the bus clock would make it hard to explore this very much.

Apparently there are ways of bypassing the cache through software. I think cat said he was doing this? Maybe it was someone else? Hard to remember.

Sphere478 wrote on 2022-02-22, 17:57:

I actually kinda had this idea, to make for lyself a unofficial k6-III+ 600EE/FAK/OVC whatever using a early k6 lid that used ink to print the logo and a laser etcher. But the early lids were slightly different than later lids and so far we haven’t conclusively discovered the exact adhesive used. I would want it to look as legit as possible for my collection :p

What I wanted to say with this:
Just because the mod was first mentioned here, that doesn't mean that it wasn't discovered earlier. Imagine someone discovered this years ago, kept the knowledge to himself and used it to make profit? Someone on Ebay is selling brandnew K6-III+/550ACR's for a while now. What if they're modded K6-2+/570ACZ? If the lid is faked properly, how can you ever know? Disturbing thought.

Sphere478 wrote on 2022-02-22, 17:57:

Yep it seems we have successfully brought k6 3+ 550 level performance/overclocks to the masses. The 570 chips are definitely the ones to do this to.

Yes, this is a big game changer.
For people like me, who aren't into collecting and only buy hardware they actually intend to use in a rig, buying a K6-III+ now has become pointless. Why should I spend 400€+ on a K6-III+/550ACR when I can have a K6-2+/570ACZ for 25€ and mod it?
Mediocre CPUs like K6-2/500AFX used to sell for 20-30€ not long ago. They're pointless now as well.

This will also have impact on the value of mainboards.
Hirsch and I won our K6XV3+/66's for 40-60€ each and they all came with a CPU. On February 8th the same board was auctioned off for 147,27€. Without a CPU. On February 21st (so just 2 days ago) another one popped up for 57€ (buy now). That one was snatched up immediately by someone else.
Grab as many Super Socket 7 boards while you still can! They're going to the moon, now that K6-III+ CPUs are so easily accessible.

RaiderOfLostVoodoo wrote on 2022-02-23, 10:16:

On February 21st (so just 2 days ago) another one popped up for 57€ (buy now). That one was snatched up immediately by someone else.

That one was up for at least about a day. I saw it on Sunday and considered getting it, it was a good price considering usual prices for SS7 boards these days. Ended up deciding against it since I already have enough SS7 boards for what I would want to use them for. Still kept an eye on it, as I was curious how fast it would sell. Saw it finally sold the next day, about the same time of day that I first saw the listing.

frudi wrote on 2022-02-23, 11:12:

That one was up for at least about a day. I saw it on Sunday and considered getting it, it was a good price considering usual prices for SS7 boards these days. Ended up deciding against it since I already have enough SS7 boards for what I would want to use them for. Still kept an eye on it, as I was curious how fast it would sell. Saw it finally sold the next day, about the same time of day that I first saw the listing.

I also did search for it on Sunday. Probably a few hours before it went online.
So it was on Ebay for roughly 24 hours.

I've been watching this board since July 2021. There were no sales between April and July.
#1 (Germany): In August one finally showed up for 20€ starting bid. It came with a 500MHz K6-2, ATI graphics (worthless), noname sound card (worthless) and a Maxtor SATA controller. I saw it only hours after it was posted, because I was searching for it every day. Send a 40€ offer which got accepted.
#2 (Germany): Another one was auctioned in September for around 45€. Came with a 450MHz K6-2. Hirsch won the auction.
#3 (Germany): In October a 2MB variant was auctioned for 20,50€. But it was broken. Randomly froze up during boot. Hirsch was the one who placed the 20€ bid. He didn't want to risk more, because we weren't sure if the issues can be fixed.
#4 (Germany): On November 29th one was auctioned for 30,50€. Came with box, but nothing else. Hirsch won the auction, but the board turned out to be broken. Turns on but no post. Fully refunded. Broken board is still in his possession. Repair attempt is planned.
#5 (Germany): On December 18th one was auctioned for 56,90€. 380MHz K6-2 included. That one also went to Hirsch.
#6 (Germany): On January 23rd an old revision (no UDMA66) was sold via buy now. Asking price was 120€. I did ask the seller if he sold it for that price or accepted a lower offer. Got no answer.
#7+#8 (Germany): On February 6th two untested ones where auctioned off for 25,75€. No extras. I won them. One turns on, but no post. The other one is fine, but I/O ports are rusty and will get replaced. Other bidders probably chickened out because the boards were untested and the seller has a bad reputation (30% negative). Lucky me! No risk, no fun!
#9 (Ukraine): On February 8th one was auctioned off for the insane price of 123,75€. No extras. With tax and shipping that would result in 170€. Auction probably went so high, because it was international. Someone in a remote location must have been really desperate.

RaiderOfLostVoodoo wrote on 2022-02-23, 10:16:

What I wanted to say with this:
Just because the mod was first mentioned here, that doesn't mean that it wasn't discovered earlier. Imagine someone discovered this years ago, kept the knowledge to himself and used it to make profit? Someone on Ebay is selling brandnew K6-III+/550ACR's for a while now. What if they're modded K6-2+/570ACZ? If the lid is faked properly, how can you ever know? Disturbing thought.

There was an ebay auction some time ago (somewhere in 2020 or in the beginning of 2021, can't remember exactly) for 50 or some more K6-III+/550ACR in original manufacturer packaging. These were NOS and as far as one can say from the pictures authentic. They were sold for ~3500 EUR back then, that's 70 EUR each. IF, and I mean IF, these are the same that are sold piecewise nowadays you can calculate the win.
I could imagine the buyer is selling them step by step now to not generate too large a supply to the market to not ruin the prices. At least I would do so.

Y'all are playing it wrong, situations like this are where you snag all the cheap 486 gear while everyone is distracted 😉

BitWrangler wrote on 2022-02-23, 14:15:

Y'all are playing it wrong, situations like this are where you snag all the cheap 486 gear while everyone is distracted 😉

shh! You're giving away secrets!

🙈🙉🙊
...anyone said something?

^ hahaha

RaiderOfLostVoodoo wrote on 2022-02-23, 10:16:

What I wanted to say with this: Just because the mod was first mentioned here, that doesn't mean that it wasn't discovered earli […]
Show full quote

Sphere478 wrote on 2022-02-22, 17:57:

I actually kinda had this idea, to make for lyself a unofficial k6-III+ 600EE/FAK/OVC whatever using a early k6 lid that used ink to print the logo and a laser etcher. But the early lids were slightly different than later lids and so far we haven’t conclusively discovered the exact adhesive used. I would want it to look as legit as possible for my collection :p

What I wanted to say with this:
Just because the mod was first mentioned here, that doesn't mean that it wasn't discovered earlier. Imagine someone discovered this years ago, kept the knowledge to himself and used it to make profit? Someone on Ebay is selling brandnew K6-III+/550ACR's for a while now. What if they're modded K6-2+/570ACZ? If the lid is faked properly, how can you ever know? Disturbing thought.

Sphere478 wrote on 2022-02-22, 17:57:

Yep it seems we have successfully brought k6 3+ 550 level performance/overclocks to the masses. The 570 chips are definitely the ones to do this to.

Yes, this is a big game changer.
For people like me, who aren't into collecting and only buy hardware they actually intend to use in a rig, buying a K6-III+ now has become pointless. Why should I spend 400€+ on a K6-III+/550ACR when I can have a K6-2+/570ACZ for 25€ and mod it?
Mediocre CPUs like K6-2/500AFX used to sell for 20-30€ not long ago. They're pointless now as well.

This will also have impact on the value of mainboards.
Hirsch and I won our K6XV3+/66's for 40-60€ each and they all came with a CPU. On February 8th the same board was auctioned off for 147,27€. Without a CPU. On February 21st (so just 2 days ago) another one popped up for 57€ (buy now). That one was snatched up immediately by someone else.
Grab as many Super Socket 7 boards while you still can! They're going to the moon, now that K6-III+ CPUs are so easily accessible.

Look at upper right corner of ceramic 😀

Did some tests today and found out that the resistor can be removed completely to unlock the cache. My 500ACZM running very well with the unlocked cache. The CPU does not need a higher vcore for this mod and does not have a higher temperature or power consumption (system).

For 600Mhz the Vcore must be increased a lot but the difference between 128kb and 256kb is low. 2.3V for 128KB and 2.4V for 256KB

Memtest

CPUID: K6-2+ 128kb lc2

1----------------------------------------------------------------
2CrystalCPUID 2.6.3.46 (C) 2002-2004 hiyohiyo
3                    Crystal Dew World [http://crystalmark.info/]
4----------------------------------------------------------------
5
6             OS : Windows 98 SE  
7           Date : 2022/02/23 13:19:30
8
9       CPU Name : AMD K6-2+
10  Vendor String : AuthenticAMD
11    Name String : AMD-K6(tm)-III Processor
12   Process Rule : 0.18 um
13       Platform : Socket 7 / Mobile
14       CPU Type : Original OEM processor
15Number(Logical) : 1
16         Family : 5
17     Generation : 6
18          Model : D
19       Stepping : 4
20        Feature : MMX 3DNow! 3DNow!+
21PowerManagement : PowerNow!/Cool'n'Quiet
22
23                    Current        Original
24          Clock :  601.35 MHz      600.00 MHz
25   System Clock :  100.22 MHz      100.00 MHz
26     System Bus :  100.22 MHz      100.00 MHz
27     Multiplier :    6.00            6.00
28      Data Rate :    SDR
29     Over Clock :    0.23 %
30
31     L1 I-Cache :   32 KB
32     L1 D-Cache :   32 KB
33       L2 Cache :  128 KB [Full:601.35 MHz]
34
35       CPUID    : EAX      EBX      ECX      EDX
36       ----------------------------------------------
37       00000000 : 00000001 68747541 444D4163 69746E65
38       00000001 : 000005D4 00000000 00000000 008021BF
39       80000000 : 80000007 00000000 00000000 00000000
40       80000001 : 000006D4 00000000 00000000 C08029BF
41       80000002 : 2D444D41 7428364B 492D296D 50204949
42       80000003 : 65636F72 726F7373 00000000 00000000
43       80000004 : 00000000 00000000 00000000 00000000
44       80000005 : 00000000 02800140 20020220 20020220
45       80000006 : 00000000 00000000 00804220 00000000
46       80000007 : 00000000 00000000 00000000 00000007
47
48       MSR      : EAX1     EDX1     EAX2     EDX2
49                : 00CAD246 00000000 00000000 00000000
50       QPF      : 1.193180 MHz

CPUID: K6-2+ 256kb l2c unlocked

1----------------------------------------------------------------
2CrystalCPUID 2.6.3.46 (C) 2002-2004 hiyohiyo
3                    Crystal Dew World [http://crystalmark.info/]
4----------------------------------------------------------------
5
6             OS : Windows 98 SE  
7           Date : 2022/02/23 10:55:55
8
9       CPU Name : AMD K6-III+
10  Vendor String : AuthenticAMD
11    Name String : AMD-K6(tm)-III Processor
12   Process Rule : 0.18 um
13       Platform : Socket 7 / Mobile
14       CPU Type : Original OEM processor
15Number(Logical) : 1
16         Family : 5
17     Generation : 6
18          Model : D
19       Stepping : 0
20        Feature : MMX 3DNow! 3DNow!+
21PowerManagement : PowerNow!/Cool'n'Quiet
22
23                    Current        Original
24          Clock :  601.36 MHz      600.00 MHz
25   System Clock :  100.23 MHz      100.00 MHz
26     System Bus :  100.23 MHz      100.00 MHz
27     Multiplier :    6.00            6.00
28      Data Rate :    SDR
29     Over Clock :    0.23 %
30
31     L1 I-Cache :   32 KB
32     L1 D-Cache :   32 KB
33       L2 Cache :  256 KB [Full:601.36 MHz]
34
35       CPUID    : EAX      EBX      ECX      EDX
36       ----------------------------------------------
37       00000000 : 00000001 68747541 444D4163 69746E65
38       00000001 : 000005D0 00000000 00000000 008021BF
39       80000000 : 80000007 00000000 00000000 00000000
40       80000001 : 000006D0 00000000 00000000 C08029BF
41       80000002 : 2D444D41 7428364B 492D296D 50204949
42       80000003 : 65636F72 726F7373 00000000 00000000
43       80000004 : 00000000 00000000 00000000 00000000
44       80000005 : 00000000 02800140 20020220 20020220
45       80000006 : 00000000 00000000 01004220 00000000
46       80000007 : 00000000 00000000 00000000 00000007
47
48       MSR      : EAX1     EDX1     EAX2     EDX2
49                : 00CAD006 00000000 00000000 00000000
50       QPF      : 1.193180 MHz