CheckEmoji Community · the emoji forum
🏠 Home 🆕 What's new ❓ Unanswered 🔥 Popular 📡 RSS Members 👥 0 online
log in · register
Home › IT › Hardware › Desktops › Overclocking (Main Thread)

Overclocking (Main Thread)

Started by dustypilot23 · · 👁 60 views · 3.3K replies

📡 Subscribe to replies

Participants dustypilot23Morgan Ortiz69John Palmer4driftingnomad68neonhawk53rustyrider40brightgull15boldhound8Bradley Myers2driftingmarlin34Aelectriceagle40placidrangeramberorca192Rebecca Parkerwiredjackal7goldenowl73Jose Cook17Tyler Lee57Alex Palmer66Grace Palmer39Scott Morris8briskheron10Thomas Ward9 …
Jason Harris2 Jason Harris2 Newcomer
8 messages
joined Nov 2006
#361 ·
I should’ve mentioned that you really need to strip away those 5 to 10 degrees that Gigabyte builds into the BIOS just as a safety net for "safe" overclocking.

As for the Democratic Party, I know a ton of people have run into issues with them, though I couldn't tell you exactly why, other than the fact that their positioning is just plain nonsensical.
My guess is that most of it comes down to crappy power supplies; personally, my rail oscillations stay right around +/-0.01v.

I'm thinking about dropping a Volcano in there pretty soon. 😁

Actually, I picked up this old Aopen tower case specifically because the side panel had this single vent designed to blow hot air directly off the processor. The reviews claimed it could drop temps by 10 degrees.

Of course, once the box actually arrived, I realized the holes weren't even there, and then I found out you can't even order the panel separately. Fujifilm, man... what a joke.👎
Carol Ortiz4 Carol Ortiz4 Active Member
57 messages
joined Sep 2003
#362 ·
Jason Harris2 said:I should have mentioned earlier that those temperature readings need to be adjusted by about 5-10°C. That’s just an offset Gigabyte builds into the BIOS as a buffer for safe overclocking.

As for the Democratic Party, I know quite a few people ran into issues with them, though I'm not entirely sure why—other than the fact that they are positioned in a rather awkward spot.
My guess is that back then, it was mostly due to subpar power supplies; personally, my rail oscillations stay within a very tight +/-0.01v range.

I think I'll be dropping a Volcano12 in there shortly. 😁

By the way, I actually went out of my way to buy this specific Aopen tower case because the side panel featured a single duct designed to direct airflow straight onto the processor to vent the heat out. The reviews claimed it could drop temps by 10°C.

Of course, when the box finally arrived, I realized the holes weren't there, and I later found out you can't order the panel separately. Fujifilm, honestly.👎

Actually, it is possible that Gigabyte is adding that offset... I haven't had the chance to take precise measurements yet... not that I really feel the need to, since it's running at roughly 2.7-2.8 GHz at about 42-52°C.

Regarding the Democratic Party... yes, it could definitely be the power supply, though some folks had trouble even with high-end units... it just goes to show how many variables are involved.

About that little tube thing... technically, it should be called a "CPU duct." Its purpose isn't to pull air away from the heatsink, but rather to funnel fresh air directly from the outside onto the CPU cooler to prevent hot air from recirculating inside the chassis.

And usually, those specific cutouts are more of a DIY project since every motherboard places the processor in a different spot, making a universal design a bit tricky. You can buy them if you want, but if you're feeling adventurous and don't want to spend the money... grab a hacksaw and start carving away at the case! 😁
Andrew Richardson2 Andrew Richardson2 Newcomer
1 message
joined Aug 2005
#363 ·
I picked this up about a year ago, and I guess I'm extremely satisfied with how it performs.

HP Workstation xw4200, dual P4 3.6 GHz processors.

Three 1GB memory sticks.

Two 160 GB drives.
Brand new 16x DVD burner.
DVD player.
A 19" flat monitor.
The Sound Blaster Audigy 2 ZS platinum pro card.
Nvidia GeForce2 Pro 512 mb ??
Windows XP Professional

Bought it about six months ago.
Just a laptop.
Gericom Cinema XXL 26660
Mobile Intel Pentium 4 Processor (2.66 GHz, 533 MHz FSB, 512KB L2 cache)
Microsoft Windows XP home edition
The ATI Mobility Radeon 9700 with 128 MB of VRAM.
60 GB hard drive.
17” TFT WXGA (1440x900)
DVD±RW - Multiformat DVD Burner
512 MB of RAM.

Any thoughts?
I won't even get into the prices. Every time I start thinking about them, I just end up with a headache.
Carol Ortiz4 Carol Ortiz4 Active Member
57 messages
joined Sep 2003
#364 ·
Andrew Richardson2 said:Nvidia GeForce2 Pro 512 mb ??

From what I recall, the GeForce 2 Pro/GTS/Ti series uses a 128-bit chip that can only theoretically support up to 128MB of memory—though, if memory serves, even the top-tier versions maxed out at 64MB.

The first commercial products hitting the 512MB mark just surfaced about a week ago, specifically looking at things like the 6800Ultra 512MB... and even those are mostly just engineering samples circulating right now.

Did you happen to catch the manufacturer on that one? I'd be curious to know who's making such a thing.
Andrew Richardson2 Andrew Richardson2 Newcomer
1 message
joined Aug 2005
#365 ·
Gerald Turner48 said:??

Maybe you just missed which manufacturer actually made it?


Sorry, man. My math was off.

Dlabs Wildcat VP990 Pro (512 MB) Graphic card

I was reading the specs from the side of the box, thinking that was exactly what I had, but I guess I was wrong.

Dlabs Wildcat VP990 Pro (512 MB) Graphic card

My old man picked this up for me from the States about a month and a half ago.

👋
dustypilot23 dustypilot23 NewcomerOP
1 message
joined Mar 2003
#366 ·
so, just curious, what kind of clock speed were you actually hitting when you overclocked it? 😁
Carol Ortiz4 Carol Ortiz4 Active Member
57 messages
joined Sep 2003
#367 ·
Andrew Richardson2 said:3Dlabs Wildcat VP990 Pro (512 MB) Graphic Card

😁 Personally, I’ve always found myself leaning toward Quadro... I just never quite felt a connection with 3Dlabs or their specific line of chipsets. It’s true that Quadro tends to command a higher price tag, but in my experience...
John Palmer4 John Palmer4 Active Member
50 messages
joined Oct 2008
#368 ·
Made in USA said:Dell Precision workstation, Intel Core i7 3.6 GHz (dual processors at 3.6GHz each)

😕 As far as I know, the Intel Core i7 3.6 GHz doesn't even support DP mode...
Grace Palmer39 Grace Palmer39 Newcomer
5 messages
joined Apr 2007
#369 ·
Just a little something I noticed today... it’s related, but...

Had some local "tech guy" swing by today—the kind who thinks he knows everything. He was trying to convince this guy to swap out everything because his PC kept crashing... get this!!!

Power supply, RAM, CPU, motherboard... the whole works.

I managed to talk him into pausing for a second and actually thinking before he replaced a perfectly fine DVD RW drive...

He handed me the machine... fifteen minutes later, he comes back to pick it up, and all those expensive components he just bought? They're sitting there completely useless.

Of course, he didn't catch on to the "why" or the "how"... and honestly, I had no intention of explaining the reason to him.

It just goes to show—not everyone is cut out for the repair business!
Olivia Lopez41 Olivia Lopez41 Active Member
59 messages
joined Oct 2003
#370 ·
Raymond Rodriguez40 said:Just a little something I noticed today... it’s related, but still...

Some local "repair guy" stopped by today to swap out parts because a client's PC kept crashing... get this!!!

He replaced the power supply, the RAM, the CPU, the motherboard... everything.

I managed to convince him to pause and actually think before he swapped out a brand new DVD RW...

He handed me the computer... then fifteen minutes later he came back to pick it up, leaving all those expensive new components sitting there unused.

Of course, he didn't grasp the "why" or the "how"... and I didn't feel like explaining it to him.

Honestly, I think a lot of people aren't cut out for repair work!

hahaha..that's definitely one of those idiots from Storm.😂 The kind of guy who can't tell a floppy cable from an HDD cable.
Grace Palmer39 Grace Palmer39 Newcomer
5 messages
joined Apr 2007
#371 ·
It didn't come straight from him, but... a few days back, there was a guy who actually showed some real talent!

Trust me... there are plenty of them here... "talents," if you want to call them that...

The only thing some of these guys can do is shove an XP disk in and call it a "repair." They don't even know the basics of BIOS settings, let alone how to use the Toolbar on the desktop!!!

It's honestly pathetic how many people at the local Church trust these guys with their computers...

And then you've got the Storm crowd—still out there peddling those absolute joke 5200 64-bit graphics cards. Best part is listening to them try to talk shop when they're explaining why it's "perfect" for someone like me, who doesn't know a thing about tech...
😁 They'll look you right in the eye while they explain why that piece of junk is exactly what you need...🙄 😉
Olivia Lopez41 Olivia Lopez41 Active Member
59 messages
joined Oct 2003
#372 ·
haha... you won't believe what this guy told me. He actually thought you could only play Doom if you bought it bundled with specific graphics cards. Then he tried to sell his buddy a floppy cable by claiming it was a hard drive cable. To top it all off, he insisted that if you put 400MHz RAM into a 333MHz motherboard, the whole thing would just fry. 😁
A Anonymous Veteran
3.6K messages
joined May 2005
#373 ·
Gerald Turner48 said:Hey,

So I recently picked up a Mobile Barton and decided to have a little fun with it...
I've got the FSB pushed to 240MHz, which is just enough to keep my RAM running at 83% of a 400MHz bus (using Kingmax... it's slow, but reliable).

The picture really says it all... 2.9GHz... and that's just on air cooling... at only 1.75V...

My only gripe is that under full load it hits about 58°C. That bugs me because I'm used to temps topping out at 49... but I'm hoping a liquid cooling setup will fix that eventually...

Check out the screenshots...



Benchmarks show it's pretty much in the same league as an A64 3400 across the board, except for memory performance.😁 😁

Sorry, I'm not trying to start a fight here, but I honestly can't wrap my head around these OC results. There are several reasons why—anyone who has ever worked with even a single mobile chip will recognize them. For one, it seems almost impossible to cool an XP-M to 2.9GHz with just air, and you'd typically need at least 2V to hit those frequencies. Plus, looking at the ClockGen readings, they show roughly 2450MHz and 205MHz (mobile owners, try tweaking the values on ClockGen without clicking "set values" to see what happens), and so on, and so forth. Of course, no two CPUs are identical, but I've seen plenty of them, and in all the international forums I frequent, I've never heard of anything like this being possible. Here's an educational article from someone who actually knows their stuff for reference.
FastLaneHW
brightgull15 brightgull15 Newcomer
5 messages
joined Apr 2005
#374 ·
John Palmer4 said:My bad, I'm not trying to start a fire here, but I just can't buy this OC result. There are plenty of reasons why—anyone who’s ever messed around with a mobile chip will see right through this. For starters, there's no way you're air-cooling an XP-M at 2.9GHz. Plus, you'd need at least 2V for a frequency like that. Even looking at the ClockGen readout, it shows about 2450MHz and 205MHz (if any mobile owners want to test this, try tweaking the values on ClockGen without hitting "set values"). And yeah, there's more to it. Sure, no two CPUs are identical, but I've seen tons of them, and on all the international forums I haunt, nobody has ever heard of anything like this being possible.

Hmm, I took a closer look at what "freak" wrote. Basically, he's saying:
"200x13.5, 2.00V, Fails Prime95, and it reached a temperature of 43C as reported by NF7Vcore and reads 40C in the BIOS, which reads from the in-socket thermal probe, that touches the back of the CPU. I upped the FSB until I reached the maximum Prime 95 stable speed, of 2640Mhz, with 2.00V."

I've had some luck myself (we're running the same NF7 V2.0 boards, though mine isn't voltage-modded, and I'm running stock NB fans, no heatsinks on the regulators, and no cooler on the SB) because my Prime95 runs fine at 200x13.5. Mind you, I only tested it for about 12 hours. I'll admit my temps were a bit higher than 43C—maybe around 50. (Oh yeah, I also have those TR SP97 heatsinks plus I cranked the Delta fan to hit 2700MHz, but I had to bump the voltage to 2.09V => the noise level was basically competing with a General Dynamics F-16 afterburner. So, I dialed the "machine" back to 200x12.5, swapped the fan for a standard 120mm, set the Enermax fans to minimum, and now it's chill.

It looks like the guy just got unlucky since his 2400FQQ4C needs 1.45V. There were some 2400 chips that ran on 1.35V, but they were incredibly rare (mostly found in European markets). Part of the issue with low overclocking potential might be this "IQZFA" thing that people rarely pay attention to when buying, even though it plays a huge role in how far a CPU can actually push. -> I think every single value from bottom-tier to top-tier has been posted over on the OC Extreme forums.
Bottom line: the fact that there's almost no difference between air cooling and an evaporator suggests the guy just didn't get a "good" batch of mobile chips. He could probably push it a little harder, though. I'm sure of it.
A Anonymous Veteran
3.6K messages
joined May 2005
#375 ·
brightgull15 said:So, I took a look at what that "enthusiast" wrote. Basically, he's saying:
"200x13.5, 2.00V, fails Prime95, hitting 43C according to the NF7 Vcore, though the BIOS shows 40C since it pulls from the in-socket thermal probe right behind the CPU. I pushed the FSB until I hit a stable max of 2640Mhz at 2.00V on Prime95."

I’m actually pretty lucky myself—we're running the same NF7 v2.0 boards, but mine isn't voltage-modded, doesn't have passive heatsinks on the voltage regulators, has a stock NB fan, and no cooler on the SB. Despite that, my Prime95 runs fine at 200x13.5. Mind you, I only stress-tested it for about 12 hours. I have to admit, my temps were a bit higher than his 43C, maybe closer to 50C. (Oh yeah, we also have those same TR SP97 passives, plus I swapped in a Delta fan just to try and hit 2700MHz. But I had to crank the voltage to 2.09V, and man, the noise level was basically competing with a General Dynamics F-16 during afterburner burn. So, I dialed the "machine" back down to 200x12.5, swapped the fan for a standard 120mm, turned the Enermax fans to minimum, and now it's solid.)

It looks like the guy didn't have much luck, since that 2400 FQQ4C needs 1.45V to run. There were some 2400 chips out there that could run at 1.35V, but they were incredibly rare (mostly seen in European markets). Part of the problem with low overclocking potential might be this "IQZFA" factor—something people don't pay enough attention to when buying, even though it dictates exactly how far a CPU can be pushed stably. I think all the various performance tiers have been posted over on the OC Extreme forums, but who knows how accurate all that really is...

Bottom line: the fact that there's barely any difference between air cooling and an evaporator suggests this guy just didn't get a "good" batch of mobile chips. He could probably push it a bit further, though; I'm fairly certain of it.

I wouldn't necessarily say he's unlucky. For a 2400 chip, that's about it—he's stable at 2640 with 2.00V, and you're hitting 2700 with 2.09V. To me, those are both decent results! I will say, though, that getting those kinds of numbers requires a bit of luck, and I agree you got a good one. Even if the NF7 helps undervolt the CPU, trying to cool a 2.09V setup with just air is pretty gutsy. The 2400 mobile at 1.45V (35w) still exists—they aren't hard to find in the US—but they aren't FQQ4C, they're FJQ4C, and they haven't really shown anything special to set them apart from the pack. The tiny gap between air cooling and Prometeia just tells me he's hitting the ceiling. Everyone hits that point eventually, as surprising as it sounds. As for the IQZFA and AQYHA, they are the most "hit or miss" series; some people get above-average results while others just get average (if you can even call a mobile series "average"). Everyone agrees the top-tier stuff is the IQYHA. Basically, this is what the relationship between frequency and voltage should look like; the kid definitely put in the work.
You mentioned you think it could go higher—if you have a link, I'd love to see it. Otherwise, if this guy (Gerald Turner48) is claiming 2900 at 1.75V, tell him to set the FSB and RAM to 1:1, run 200x14 at that voltage, and post some screenshots of the CPU Arithmetic and multimedia benchmarks along with SuperPi. I'm genuinely curious (though, forgive my skepticism, because I doubt it'll even boot with those settings). Legend has it there was only one mobile processor capable of being stable at 2700 with 1.7V, and some guy sold it on eBay for a massive profit. Maybe Gerald Turner48 is that lucky winner, but I highly doubt it...
Clock speed vs core I'm repeating myself here, but no two CPUs are identical. This is just a rough idea...
Cheers
brightgull15 brightgull15 Newcomer
5 messages
joined Apr 2005
#376 ·
I gave it a shot, but I’m having zero luck with 200x14 or even 2.1x V. It won't even boot DOS—thanks for nothing, really (just a bunch of "never seen this before" errors). Honestly, I think 2700 is the absolute ceiling for what this Thermalright SP97 can actually cool. And who knows, maybe my 430W Enermax is screwing me over by dropping voltage under a full load, or maybe it's something else entirely. Long story short, 2700 is the max I could squeeze out without even touching the FSB above 200. (Keep in mind we're talking about the 2500 version of the Mobile Barton, and I suspect it's an IQYnešto—but I'm too lazy to strip down the rig just to run SP97 and check.)
Carol Ortiz4 Carol Ortiz4 Active Member
57 messages
joined Sep 2003
#377 ·
marg said:and yeah, the gauges on ClockGen show about 2450MHz and roughly 205MHz (if any Mobile Barton owners want to try tweaking the values in ClockGen, just remember you don't actually have to click "set values")

If things are working for me, I really don't feel the need to defend myself, do I? Why would I start posting nonsense? It isn't my life's mission to spend all day trying to convince people of my supposed "god-tier" overclocking skills.😁

By the way, there is really only one reason I use ClockGen. When I crank the FSB up to 200MHz (or higher), EasyTune tells me the PCI is at 66MHz and the AGP is at 100MHz... I was basically just using ClockGen to verify that the buses were staying "correct."

Unfortunately, I had to scale back to 220*12, which is roughly 2650MHz... because things were getting pretty toasted (hitting about 68°C) and I wasn't particularly keen on taking that kind of risk. Now it's running at that frequency with a max temp of around 52°C...

The strange part was that MemTest86 would crash after about half an hour (at 2880), even though everything seemed perfectly fine within Windows...

I did manage to get the processor to run at 231*13, if I recall correctly... which is just a hair over 3000MHz. However, the VCore was sitting around 2.1V, and I was using a water cooler at the time (I went over to a friend's place to test it out, since he has the setup and I wanted to see how good this CPU actually was).

It seems I actually ended up with a beast of a processor, because when I tried the exact same chip at my friend's place (he picked up his mobile unit locally... I believe from Ve-mil), it wouldn't go past 2.5GHz even at about 1.8V...

That’s why I always say: stick to American products.
Carol Ortiz4 Carol Ortiz4 Active Member
57 messages
joined Sep 2003
#378 ·
Regarding the power supply, I actually had to swap it out because my old unit just couldn't handle pushing things past 2.5GHz... I've moved over to a Tagan 480 w now, and honestly, I'm quite pleased with it...

As for the memory situation: trying an FSB 1:1 ratio just wasn't an option for me... I'm running some Kingmax 433, and it refuses to cooperate even at 434MHz with the VDimm set to 2.8V... so, I've settled on frequencies where I can keep the memory stable within the 380-420MHz range...

When it comes to the motherboard itself, the MOSFETs are being cooled pretty aggressively, much like the Northbridge, Southbridge, and the memory... internal case temperatures are hovering right around ambient + 2°C...

Regarding that "code," I might take a look tomorrow... I actually just got this rig back from a friend yesterday (it was running a water cooler), so I'm not exactly itching to tear down that Vulcan setup constantly... especially considering how many capacitors are crowded around the socket 😁
A Anonymous Veteran
3.6K messages
joined May 2005
#379 ·
Tommy said:Regarding the power supply, I had to swap it out because my old one couldn't push me past 2.5GHz... now I'm running a Corsair H100i and I'm pretty happy with it...

As for the RAM: trying a 1:1 FSB ratio just wasn't happening... I've got some Kingmax 433 sticks that won't even hit 434MHz even with the voltage bumped to 2.8V... so I had to settle for frequencies where I can keep the memory running in the 380-420MHz range...

When it comes to the motherboard itself, the Intel chips are being cooled aggressively, just like the Northbridge, Southbridge, and the RAM... internal case temps are staying right around ambient + 2°C...

About those "codes," I might take a look tomorrow... I actually just got this rig back from a buddy yesterday (it's on water cooling), so I really don't feel like tearing down this Vulcan setup constantly... especially with all those capacitors crowded around the socket 😁

Kudos to you if that's working for you—honestly, seeing any solid OC results makes my day. But like I mentioned, while the numbers look amazing, there are just too many weird variables involved, so please don't take it personally when I express some skepticism. On an NVIDIA board, if I recall correctly, once you set a 6:5 ratio in the BIOS, the arrow on the clockgen turns gray (meaning you can't use it), which isn't necessarily a bad thing since we know NVIDIA needs to stay at 1:1, and any serious overclocking is obviously done strictly within the BIOS.
If your system is crashing after thirty minutes, that’s not surprising at all. For your board, much like other NVIDIA setups, things aren't truly stable until they've survived 10-20 hours of MemTest86 and Prime95, no matter how flawless everything seems while you're just hanging out in Windows.
You mentioned hitting 3003 MHz (231*13) using 2.1v, but earlier you were at 2886 MHz on just 1.75v... sorry, but that sounds a little funky, doesn't it?

You said:
“As for the RAM: trying a 1:1 FSB ratio just wasn't happening... I've got some Kingmax 433 sticks that won't even hit 434MHz even with the voltage bumped to 2.8V... so I had to settle for frequencies where I can keep the memory running in the 380-420MHz range...”

At 200, that Kingmax should be rock stable, right?
I was just asking for a 1:1 ratio at 200, which should be a breeze. Like I said, if you don't mind, try setting it to 1:1 at 200x14 with 1.75v and send over some screenshots of the CPU Arithmetic and multimedia benchmarks along with Prime95—I'm dying to see them....
And BTW, here in the USA, you run into all sorts of stuff; you just need a bit of luck and to know where you're shopping, since luckily most things can be returned...
Cheers, and I'm looking forward to those screenshots...

Oh, and tell your buddy not to be afraid of bumping up the CPU voltage a bit more. What's 1.8v anyway? Especially with high-end water cooling, you see how Yahoo handles 2.1v without breaking a sweat—that's the true spirit of overclocking... 👍
brightgull15 brightgull15 Newcomer
5 messages
joined Apr 2005
#380 ·
stormyorca6 said:Yeah, tell your buddy he shouldn't be scared to crank up that voltage. It’s just 1.8V—especially if you've got a solid liquid cooling setup. Just look at how brightgull15 pulls off 2.1V without breaking a sweat. That's what a real OC enthusiast looks like... 👍

Yeah, sure, whatever you say. That was a total fluke, and even then, it was only while I was running Prime just to see if the thing would actually hold. While it was running, I was basically praying to every saint in the book that my processor wouldn't fry. Now, it's sitting pretty at a stable 2500 (and it's quiet, too).

You must log in or register to reply here.

Log in Register

🔗 Similar threads