Sure there are, even 400MHz CL2 : LINK.
They need manual tweaks (timings + voltage), to work properly at those settings though.

I would check how E7200 behaves when locked to 800MHz FSB.
Also, I'm not sure if your memory "likes" CL2.5 at 400MHz, maybe they have to be overvoltaged a bit to work (to 2,5V, for example) ?

For this board, its recommended to use Pentium E5x00 CPUs, most preferably E5800.

agent_x007 wrote:

Sure there are, even 400MHz CL2 : LINK. They need manual tweaks (timings + voltage), to work properly at those settings though. […]
Show full quote

Sure there are, even 400MHz CL2 : LINK.
They need manual tweaks (timings + voltage), to work properly at those settings though.

I would check how E7200 behaves when locked to 800MHz FSB.
Also, I'm not sure if your memory "likes" CL2.5 at 400MHz, maybe they have to be overvoltaged a bit to work (to 2,5V, for example) ?

Thanks for the feedback, I do appreciate it.

Ahh, such RAM timings do exist, however they're sold out in that link, so maybe they did exist, like the dinosaurs... 😀

I did try to over-volt to 2.6 and 2.7V. It is an option in the BIOS, but that didn't work, sadly. None of my DIMMs have heat-spreaders. The fancy DIMMS that claim 400@CL2.5 have them. I can't tell if those are cosmetic or are necessary. Do manufacturers pick cool looking heat spreaders at higher cost for better marketing appeal, or does the commodity nature of the market drive them away if unnecessary? None of my DDR DIMMs have ever gotten hot, so I don't know.

I don't know how to lock the chips to FSB800 -- might not be possible with E7200. Any suggestions will be appreciated.

On the bright side, I did experiment with the clocking on these chips. On the Celeron 420, I was able to drop the base clock to 100MHz from 200MHz. I tried 95Mhz, but things got unstable, as can be expected due to failing setup and hold times. I then was able to change the BIOS setting for "On-demand Clock Modulation" to 1/8th and 1/4th. This means that the 1.6GHz clock is first dropped to 800MHz with the base clock change, and then the clock modulation change gave me effective performance of 100MHz and 200MHz, respectively. To test this, I ran Descent 1 in Windows98SE. The 1/8th * 800 setting gave me slow pentium and 486DX2 performance from Quake, Chasm, and Descent. The 1/4 performance gave me performance about on par with P200 MMX. This platform gives flexibility that may only be matched by the Super Socket 7, but in a completely different range.

Chips that support Speedstep can specifically pick their multiplier down to about 1/2, so Wolfdale and Conroe chips have even greater flexibility. A chip with small multiplier but high FSB has the most flexibility, as long as you operate in ratios that aren't too far from what the RAM and FSB need to meet timing.

Here's the power meter on the system using integrated graphics. Descent was running right at intended speed, which was surprising since I haven't been able to make a system do that correctly since my socket 7 pentium with clock multiplier jumpers and cache toggle.

Chips that support Speedstep can specifically pick their multiplier down to about 1/2, so Wolfdale and Conroe chips have even greater flexibility. A chip with small multiplier but high FSB has the most flexibility, as long as you operate in ratios that aren't too far from what the RAM and FSB need to meet timing.

SpeedStep really doesn't work like that.
ALL CPUs based on Core 2 can use x6 multiplier (when SpeedStep is disabled), regardless if locked or not. So it's exact opposite of what you described.
In reality lowest FSB with highest multi get's you best flexibility.
How to set locked FSB ?
You go to BIOS, set FSB to 200MHz and save settings.

This isn't rocket science you know...
Also, make sure primary graphics slot is "AGP/PCI" not "PCI/AGP".
PS. Memory I used :

agent_x007 wrote:

SpeedStep really doesn't work like that. ALL CPUs based on Core 2 can use x6 multiplier (when SpeedStep is disabled), regardless […]
Show full quote

Chips that support Speedstep can specifically pick their multiplier down to about 1/2, so Wolfdale and Conroe chips have even greater flexibility. A chip with small multiplier but high FSB has the most flexibility, as long as you operate in ratios that aren't too far from what the RAM and FSB need to meet timing.

SpeedStep really doesn't work like that.
ALL CPUs based on Core 2 can use x6 multiplier (when SpeedStep is disabled), regardless if locked or not. So it's exact opposite of what you described. In reality lowest FSB with highest multi get's you best flexibility).
You go to BIOS, set FSB to 200MHz and save settings.

This isn't rocket science you know...
Also, make sure primary graphics slot is "AGP/PCI" not "PCI/AGP".

If 6 is the lowest multiplier for Core 2 chips, and the ceiling for Core 2 clocks is around ~3.2 GHz, that means that FSB * multiplier will never be greater than ~3.2 GHz.

base_clock * 4 = FSB

FSB1066:
266 * 12 = 3192
100 * 6 = 600
Freq range = 3192 - 600 = 2592

FSB800:
200 * 12 = 2400
100 * 6 = 600
Freq range = 2400 - 600 = 1800

2592 > 1800 == true

Am I missing something?

Anyway, setting the base clock for the E7200 to 200 doesn't work. The machine won't POST at all, which makes me wonder if the (*) advisory on the supported CPU list isn't just a warning -- it is also a requirement.

The DDR400 CL2.5 sticks showed up, and alas, they do not work with the E7200. I sent Asrock an email to maybe clarify which model, if any, of E7200 that works with this board. It may save someone else some time in the future.

E5800 is on the way. Only $5!

"Ceiling" for Core 2 chips IS A LOT higher than 3,2GHz.
However, because this board doesn't support high FSB speeds (or Vcore adjustment), you are quite limited in what you can do (to maximum multiplier and maximum FSB).
In case of my E6500(K) I could set 3,7GHz :

While my E7500 went only to 3,3GHz :

I used Hirensboot MiniXP to get the screenshot (that's why only one core is used).

PS. Do you have a PCI graphics card ?

No fair! Unlocked multiplier! 😀

Daaang that's cool. Are you able to drop the multiplier below 6 on the 6500k? Any special cooling requirements?

Cooling : Intel BOX "High profile" 😀
For 3,7GHz you need 18,5 multi (for 800MHz FSB) and 14 multi for 1066MHz.
Sadly no. But 573MHz is too high for you (there is that modulation you can do as well) ?

Also, E7500 at 600MHz :

AND best CPU for this board :

Went to 3,2GHz (but not sure if it's stable)

I'm just wondering if the unlocked multiplier chip can get lower than 6. The reason is to be able to use the clock modulation option to slow down to 486 effective performance for speed sensitive games. Not a requirement, but a nice ability. I'm moving soon and have started putting lots of stuff into storage, so consolidation is desirable.

It can't go lower.

agent_x007 wrote:

It can't go lower.

Understood. I've already gotten my expectations a little too high considering the surprising ability to divide 600/8 and get (effective) 75MHz . I have not been able to get any other chip since my Socket 7 (P200MMX and K6-2 266) down using multipliers. I haven't come across anything (Slot 1 all the way to 939) that has this kind of flexibility.

I bought an Allendale back when it was brand new and I overclocked it from 2.0GHz to 3.0GHz with stock cooling and perferct reliability. The machine still runs to this day. It was the Celeron 300A of 2007. Shortly after that I quit using computers for 5 years, so I never saw where 775 ended up.

Thanks for posting the responses and photos, agent_x007. I appreciate the feedback.

No problem.
Here's how my OC'ing of QX6850 turned out :

In short : Basicly, a QX6800 "G0" with +1 on multiplier.

MOSFETs used on my board are rated at 55A @ 100C.
They can do up to 75Amps (if cooled enough).
I say it's pretty good OC, considering I used 120mm fan on top of BOX cooler (and I added heatsinks to all inductors on top of that, to help with VRM heat) 😀

PS. I DO NOT RECOMMEND using Quad Cores on this board (unless you have a fan forcing cool air over VRM).

Just found this thread.
I've been having exactly same issues, but with r2.03 version of the board and E5800[SLGTG/R0] / E5700 [SLGTH/R0]. Board loads correct microcode and all that.
I suspect it has something to do with cpu-chipset-ram mix (ram is Patriot DDR400 CL3 1GB sticks). Sometimes it posts with only one stick, most of the time it does not. But when it posts, it is successful every time after that ... until complete power off.

Tested before with Celeron 351/533 FSB, no problems. Tested again today with E4500/800 FSB. Works every time with no issues. I even set RAM to CL2.5.
Overclocked CPU to 250MHz FSB, with RAM on 5/4 ratio (effectively 400MHz and CL2.5) ... also no problems.

I have incoming some higher quality/lower latency RAM kits so I'll be testing it again.

So, here's the deal: it's down to the specific board!
Got 2nd board (also r2.03) and some CL2 Infineon BE-5/Brainpower PCB memory sticks. Everything that doesn't work on 1st board, works on 2nd. Well, sort of ... it does sometimes take a bit longer to boot, and sometimes I have to press power button few times for it to start correctly. But when it starts, it's all OK from that point forward. No such issues with any other supported CPU I tried.
To conclude: support for Wolfdale on r2.03 is sort of wonky ... it may completely work, or not work with various states in between. I guess ASRock didn't put CPUs on the list for good reason.

Also, although boards are of same revision, they are not completely identical. 2nd one uses solid capacitor on top of the socket, while 1st uses regular one. 1st board also ran a bit warmer.

Ehm... According to ASRock, R2.x boards do NOT support Wolfdale CPUs: https://www.asrock.com/mb/Intel/775i65G%20R2. … ndex.us.asp#CPU
So the fact that you are able to at least (sometimes) boot with a Wolfdale CPU is quite impressive 😀
I am using an E5800 CPU on an R3.0 board and it's working perfectly (but this one officially supports Wolfdale CPUs). I also have an R2.12 board with a Core2 Duo E4700, which also works perfectly.
My advice is to stick to Conroe on R2.x revisions. It's already an extreme overkill for Win98 builds. Most games will run at 500 or even 1000+ FPS. The Quake 2 timedemo, in software mode, at 1024x768, scores 120 FPS. What more do you need? 😁

I know 😁. I had hope after running into this thread:
Asrock 775i65G R3.0 vs. R2.03

Anyway, C2D E4x00 it is. BTW, I don't see many E4700s on used market.

Great! 😀 You can even overclock that bad boy to 3 - 3.2 GHz and you'll have one of the fastest Win98 PCs out there.
Here's one, £3.50 + shipping: https://www.ebay.co.uk/itm/Intel-Core-2-Duo-E … YUAAOSwDyReTU1V

Hi Everyone,
Kind of late to the party but I have the v2 board and have made a few observations:

1. Upgrade the BIOS immediately. The latest 3.x BIOS works on the 2.x boards. I see screenshots in this thread of a 1.0 BIOS that made me kinda shiver. I couldn't get my 2.x board to post with an E7500 until I updated the BIOS. It started with 2.03. I had to actually put a Pentium D 930 in it to get it to boot. I was able to flash it with the latest P3.30 and everything started working.

2. Very unreliable boot up with 1066 FSB chips. This board seems to work fabulously with any 775 chip with a 800 bus speed (or so I thought). The 1066 ones is where things get dicey. I think to wrap your mind around it, and have predictable results with this board, one has to put themselves in the time of when it was created. This board has 1066 stamped on it, but actually this board is *overclocking* itself to run at those speeds. The problem is, it's actually not really stable at 1066 (from my experience). It defaults to a 266 base clock for a 1066 CPU and that's where things just don't work or are extremely unstable. I was able to manually lower it to 233 and everything got stable. For reference, the base clock is 200 for an 800 FSB. I think it's got to do with how well the old DDR400 memory can handle over-clocking. The board's AUTO setting for base clock is too high and leads to instability with the default settings. I've seen it do it's "startup stammer" where it takes a while to even post, seeming to figure out what the timings of everything should be-- and it sometimes would like half-clock the CPU. It'd startup at like 1.1 Ghz instead of 2.0. Which leads to...

3. The memory you use makes a BIG DIFFERENCE with this board. Get a nice pair of CL2.5 DDR400 chips, the kind you can overlock, with heat spreaders. This board wants to work. It will try to work with almost anything I think you can throw at it. It just takes forever to post, and you'll see crazy timings for your memory and CPU. Remember this board is overclocked by default with 1066 chips. My road to self-discovery started after trying numerous 1066 chips without satisfaction.

I eventually arrived at the same conclusion someone else stated above about the E5800 CPU. It's a fast 3.2 GHz chip, but with an 800 bus. I thought the board just couldn't run reliably at over 800. So I found an E5800. It seemed like it would make all my dreams come true for this board-- that is, be something that would be fast and reliable without a bunch of tweaking. But it was not to be, or so I thought. Being that all other 800 bus chips worked flawlessly without compliant, I confidently installed it and turned on the system...and nothing. Nothing! It wouldn't even try to post. Only once did it stammer and post, allowing me to peek into the BIOS. It was showing that it was clocked below 1 GHz! I was confused, what could it be? Could it be that the E5800 was too new and had some microcode that the BIOS just couldn't figure out what to with?

I was stumped and defeated. The E5800 seemed to have a bigger problem. So I turned my attention back to the Wolfdale E6700 that worked fairly well, ironically (it's the same chip just different FSB speed). it just isn't stable at boot time (it'd take 30 seconds to post, when it did post). I began to theorize that maybe my problem with it was the RAM. I had a pair of CL2.5 DDR400 chips like the manual suggested. But they just looked wimpy. I mean they are just bare chip sticks, something you'd find in a low-end Dell office computer. So I got back on eBay and sleuthed around, honestly, just looking for memory that "looked like memory someone could overclock." I was making a hail marry. So I found a pair of 1GB DDR400 chips that could do CL2.5, but this time they looked like something you'd buy to overclock... they had the heavy metal heat spreaders. Then this is when things started to come together...

I put the new memory in and the E6700 worked better. It still couldn't run it at full speed but it seemed to post a lot more reliably. Okay, this is good I thought. I wondered what the E5800 would do? So I reinstalled it. What do you know? It posted right away running at full speed, 3.2 GHz. BIOS acted like it knew what to do. Everything just worked. And it was stable. No overlocked FSB, runs at full 3.2Ghz with 16x multiplier.

So now it's running stable with the following config:

P3.30 BIOS
2GB (2x1GB) CL2.5 DDR400 memory (with head spreaders)
E5800 Wolfdale 3.2 Ghz / 800 FSB / 2 MB cache with Zalman cooler
ATI Radeon 9800 XT - replaced cooler with a Zalman cooler (super quite and cooler now)
SATA 500 GB Seagate drive sized down to 32 GB with Seatools
Windows ME

It scores 31404 on 3dMark2000 and 22556 on 3dMark2001!

Side note:
I learned more about how Intel named the Core 2 CPUs than I ever thought I'd need to. Reading Wikipedia's article about Core2 chips, I got a totally wrong idea about how things worked with the numbering scheme. The article actually has quite a few errors in it, especially it seems with the newer chips. I was thinking that the E5xxx/E7xxx came first and where older, and replaced by the E6xxx/E8xxx later. That wasn't correct at all, 🤣. Turns out the E5800 was one of the last "Core 2" chips released. It is a 45nm chip and is actually a Wolfdale (like the E6700). The thing was not marketed as a Core 2 though-- it was sold as a Pentium and released Q4 2010!.

But it gets even worse! There are *two* E6700 chips!!! Both are 1066 bus, but have different cache and speed! The first was a Conroe released in Q3'06 with 4M cache at 2.66 GHz. It's called "Core 2 E6700." The other is a Wolfdale released in Q2'10 with 2M cache at 3.2 GHz! It's called "Pentium E6700." So it seems Intel released die-shrinks of chips or reused model numbers (partially).

https://ark.intel.com/content/www/us/en/ark/p … 66-mhz-fsb.html
https://ark.intel.com/content/www/us/en/ark/p … z-1066-fsb.html

Sadly no possibility to change Vcore as Win9x doesnt seem to support speedstep 🙁