-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
Does any KT333 have a /5 divisor?
- Thread starter Falceron
- Start date
The EPOX boards apparently do - and just to avoid confusion the BIOS dated 28 MAR 02 now displays the AGP and PCI bus speeds below the FSB in the BIOS, removing any doubt.
Of course, I just got my board today, and haven't fired it up yet, so I can't verify that yet
Of course, I just got my board today, and haven't fired it up yet, so I can't verify that yet
OP
- Joined
- Aug 24, 2001
- Thread Starter
- #3
While waiting for a response I asked on the hardocp forums heh.. I'll probably end up getting the 8k3a+ since it seems like such a winner. Is this limited to the epox board, or do other KT333 based boards have it too?
What's your setup, vandersl?
What's your setup, vandersl?
My setup is pretty much what is in my sig - only with the 8K3A+ and 512MB of Samsung PC2700. CPU upgrade will come when the new T-Breds come out.
After installing the board and reinstalling windows once RAID0 was set up, I tweaked just a little bit. So far, the board and memory are stable at 175MHz FSB, Cas2, 2-5-2, 1T command. It got too late last night to try any further, but I'll let you know after this evening.
The board definitely supports 1/5 PCI and 2/5 AGP dividers - or the BIOS is lying to me.
So far, I am mighty impressed with the memory and board.
After installing the board and reinstalling windows once RAID0 was set up, I tweaked just a little bit. So far, the board and memory are stable at 175MHz FSB, Cas2, 2-5-2, 1T command. It got too late last night to try any further, but I'll let you know after this evening.
The board definitely supports 1/5 PCI and 2/5 AGP dividers - or the BIOS is lying to me.
So far, I am mighty impressed with the memory and board.
OP
- Joined
- Aug 24, 2001
- Thread Starter
- #5
You seem to be planning almost exactly the same system that I am. Differences would be only hard drive and CPU cooler. I'm planning on getting the new Alpha, but that may change by the time I upgrade. So you have a single 512 meg stick of Samsung PC2700? Where did you pick that guy up? The 256 version was supposed to get incredible results. Check out some posts on Hardocp about it if you don't frequent there. Let me know how everything goes!
Actually, I have two 256MB sticks of Samsung PC2700. I had a 512MB stick of the Corsair PC3000 in my shopping basket on Sunday night, but on Monday morning when I went to place the order, they were out of stock, so I went with the Samsung. Saved a few bucks anyway.
A quick update - I have tested everything out at a FSB/mem of 180 at the fastest settings (2-2-5-2, 1T). Everything is working fine - ran Prime95 for 4 hours no problem, and multiple loops of 3DM2K1.
I am a little concerned that the BIOS is lying to me. Sandra is reporting my AGP and PCI speeds as 90MHz and 45MHz, respectively, though the BIOS shows 71MHz and 36MHz. However, Sandra doesn't seem to recognize the chipset, and listed my memory incorrectly at 133MHz when I was running 133/166 initially, so who knows.
All I really know is I can get Sandra memory speeds of 2400MB/s+ with a regular old Athlon TB, so I am pretty happy.
A quick update - I have tested everything out at a FSB/mem of 180 at the fastest settings (2-2-5-2, 1T). Everything is working fine - ran Prime95 for 4 hours no problem, and multiple loops of 3DM2K1.
I am a little concerned that the BIOS is lying to me. Sandra is reporting my AGP and PCI speeds as 90MHz and 45MHz, respectively, though the BIOS shows 71MHz and 36MHz. However, Sandra doesn't seem to recognize the chipset, and listed my memory incorrectly at 133MHz when I was running 133/166 initially, so who knows.
All I really know is I can get Sandra memory speeds of 2400MB/s+ with a regular old Athlon TB, so I am pretty happy.
OP
- Joined
- Aug 24, 2001
- Thread Starter
- #7
Ah yeah.. Personally I'm not a big fan of Sandra.. Or at least I don't put that much faith in its numbers I would be more apt to believe that Sandra is misinterpreting something than that Epox would deliberately lie about the information they just added to the BIOS. So is 180 the best you can do with those timings or is it just as far as you've tested right now?
I would be more apt to believe that Sandra is misinterpreting something than that Epox would deliberately lie about the information they just added to the BIOS. So is 180 the best you can do with those timings or is it just as far as you've tested right now?
- Joined
- Oct 4, 2001
- Location
- Milton Keynes, U.K.
SPELLING! It's DIVIDER NOT DIVISOR. AARRRRRRRGGGGHHHHHHHHH
Thanks for making such a useful contribution Maximus
Actually, divisor is the right term. From the dictionary (a big book with words in it):
divisor - The quantity by which another quantity, the dividend, is to be divided.
So, in this case, 5 is the divisor, and the FSB is the dividend. The circuit that actually divides the FSB by the divisor is the divider.
Actually, divisor is the right term. From the dictionary (a big book with words in it):
divisor - The quantity by which another quantity, the dividend, is to be divided.
So, in this case, 5 is the divisor, and the FSB is the dividend. The circuit that actually divides the FSB by the divisor is the divider.
To get back on topic - 180 is the highest that I have actually tested for stability.
I actually had it up and booted at 200MHz with aggressive timings (I didn't think it was going to boot), and got Sandra memory scores for it, but after a bit it crashed in PC Mark 2002.
It crashed rather badly, and the registry got corrupted on the next boot, so I did a fresh install of windows and now have a ghost image. I will be pushing the limits further now that I have a safety net
I actually had it up and booted at 200MHz with aggressive timings (I didn't think it was going to boot), and got Sandra memory scores for it, but after a bit it crashed in PC Mark 2002.
It crashed rather badly, and the registry got corrupted on the next boot, so I did a fresh install of windows and now have a ghost image. I will be pushing the limits further now that I have a safety net
OP
- Joined
- Aug 24, 2001
- Thread Starter
- #11
Heh good idea I recently re-partitioned to set up a 3 gig windows partition that I can image. That should help so much if I manage to screw something up. Know of any decent places that stock that 512 Samsung module? I see Newegg has the 256 stick available, but that seems to be all.
I recently re-partitioned to set up a 3 gig windows partition that I can image. That should help so much if I manage to screw something up. Know of any decent places that stock that 512 Samsung module? I see Newegg has the 256 stick available, but that seems to be all.
- Joined
- Oct 4, 2001
- Location
- Milton Keynes, U.K.
vandersl said:Thanks for making such a useful contribution Maximus
Actually, divisor is the right term. From the dictionary (a big book with words in it):
divisor - The quantity by which another quantity, the dividend, is to be divided.
So, in this case, 5 is the divisor, and the FSB is the dividend. The circuit that actually divides the FSB by the divisor is the divider.
lol And your contribution??????
In PC terms we call the division of the PCI bus speed through 1/3 1/4 1/5 1/6 Dividers, the term divisors is certainly not as widespread, espicially in the UK.
M_N
Well, I may not have 11.67 posts per day, if you want to call that a contribution (I especially like that last one suggesting the person lie to get an RMA). But at least I don't post needless troll bait (other than responding to this one).
As if you didn't know what '/5 divisor' meant, and posted that response for any reason other than being an a**. Especially in light of the fact that '/5 divisor' is correct, even using the Queens English.
As if you didn't know what '/5 divisor' meant, and posted that response for any reason other than being an a**. Especially in light of the fact that '/5 divisor' is correct, even using the Queens English.
- Joined
- Aug 3, 2001
- Location
- VARIANTE ASCARI
The reason VIA came up with the KT333 chipset is to have support for 1/5 & 1/6 divisors. This is probably why you don't get too many replies to this thread from the more experienced members.
I have a question for the KT333 owners or would be owners. The reason why we overclock thru the FSB is to speed up the whole system. Why would you want to have these higher divisors that just makes your PCI & AGP to run asychronous of your CPU & Northbridge? You loose the performance increase & with CPUs that will not take full advantage of the higher BUS speeds the performance gains are minimal. Your going to need memory that will significantly be faster than DDR400 which will be available but in the newer DDRII sticks which will not fit you MoBos DIMM Slots.
I have a question for the KT333 owners or would be owners. The reason why we overclock thru the FSB is to speed up the whole system. Why would you want to have these higher divisors that just makes your PCI & AGP to run asychronous of your CPU & Northbridge? You loose the performance increase & with CPUs that will not take full advantage of the higher BUS speeds the performance gains are minimal. Your going to need memory that will significantly be faster than DDR400 which will be available but in the newer DDRII sticks which will not fit you MoBos DIMM Slots.
sonny, the reason is memory bandwidth. Take for instance 2 athlons, 1 running at 12X100 and the other running at 9X133. The Athlon running at 9X133 will be slightly faster than the 12X100 athlon, even though the PCI and AGP are both running at exactly the same speeds. the difference will be more noticeable in multimedia applications.
I just purchased an Epox 8K3A+ and XP1600+. I plan to run at 10.5X166=1743 MHz. At these speeds, my PCI and AGP will be in spec, and only the processor and memory will be overclocked. (and, I don't have to unlock the chip!)
Also, there is plenty of memory available now that will reach 170 FSB at aggressive timings, or 180+FSB with slightly less aggressive timings. So, no worries there, either.
I just purchased an Epox 8K3A+ and XP1600+. I plan to run at 10.5X166=1743 MHz. At these speeds, my PCI and AGP will be in spec, and only the processor and memory will be overclocked. (and, I don't have to unlock the chip!)
Also, there is plenty of memory available now that will reach 170 FSB at aggressive timings, or 180+FSB with slightly less aggressive timings. So, no worries there, either.
- Joined
- Aug 3, 2001
- Location
- VARIANTE ASCARI
Yes we all know bandwidth is important & I acknowledged that in my first reply to this thread. 100MHz is outdated & 133MHz so they can not be used as an example since chips these days will comfortably do them. What you need to take a look at is going over spec like in 166MHz - 200MHz. Memory maybe available but the quality sure isnt. The fact is that the TB or even the XPs may reach those FSB speeds but are they using it efficiently LOOK HERE at the benchmarks. When the divisors kick in you actually lose out. What I would like to see is what the difference is in performance in the higher FSB settings. I will be starting a thread on this as soon as I figure out what topic it should be posted at
LOOK HERE at the benchmarks. When the divisors kick in you actually lose out. What I would like to see is what the difference is in performance in the higher FSB settings. I will be starting a thread on this as soon as I figure out what topic it should be posted at
Last edited:
ok, here's the test I propose. When I get my 8K3A+, I'll stick an unlocked Tbird 1.4 in it. I'll run one set of benches at 10.5X134=1407 MHz and anothe set at 8.5X166=1411 MHz. (assuming my crucial pc2100 can hit 166)
The PCI and AGP should be in spec at 166, and only slightly overclocked at 134. There is only 4 MHz difference in processor speed.
Here's what I expect. CPU benchmarks will be very similar. Memory bandwidth should be noticeably different. 3Dmark will probably show some difference.
Then again, I may have to eat my words. It wouldn't be the first time.
The PCI and AGP should be in spec at 166, and only slightly overclocked at 134. There is only 4 MHz difference in processor speed.
Here's what I expect. CPU benchmarks will be very similar. Memory bandwidth should be noticeably different. 3Dmark will probably show some difference.
Then again, I may have to eat my words. It wouldn't be the first time.
- Joined
- Aug 3, 2001
- Location
- VARIANTE ASCARI
- Joined
- Oct 4, 2001
- Location
- Milton Keynes, U.K.
vandersl said:
Okay,
First I'm correct, if you bothered to read my post I said that DIVIDERS is more commonly used. I didn't mention you're term divisors.
And as for calling me an a** who posts rubbish (implied) well then you have a lot to learn, you proved by arguing that you are a small minded individual who's here for a debate.
If you want to take a spare minute to look around you and compose yourself then do it, and get out.
As for implying that I spam and post rubbish well you've really crossed the line, I've helped more people here than you can dream of. And until you realise that you won't be benefiting from any help from me.
Consider yourself, on my "ignore list" I have better things to do than argue with people like you.
Last edited:
Hi again Maxi,
Just to clarify, my initial issue was with your first post which was loud, non-constructive, and (incorrectly) criticised the spelling of 'divisor' (which was only ever used in the title of the first post, and never actually by me).
Instead of being helpful, you exhibited a 'small mind' by feeling the need to make that post, when it really didn't serve any purpose. Maybe you were just venting after a bad day - I don't know. But that is what I meant by 'contribution' in my response - you made no contribution to this thread, which is what is relevant.
Just to avoid falling into the same pit myself:
Sonny - I will be contributing to your AGP/PCI/FSB ratio)) thread later on tonight, once I get home and can run some tests. Unfortunately my video card doesn't like going to 83MHz AGP, so I can't do the best test, which would be at the 165/166MHz crossover point.
Just to clarify, my initial issue was with your first post which was loud, non-constructive, and (incorrectly) criticised the spelling of 'divisor' (which was only ever used in the title of the first post, and never actually by me).
Instead of being helpful, you exhibited a 'small mind' by feeling the need to make that post, when it really didn't serve any purpose. Maybe you were just venting after a bad day - I don't know. But that is what I meant by 'contribution' in my response - you made no contribution to this thread, which is what is relevant.
Just to avoid falling into the same pit myself:
Sonny - I will be contributing to your AGP/PCI/FSB ratio
)) thread later on tonight, once I get home and can run some tests. Unfortunately my video card doesn't like going to 83MHz AGP, so I can't do the best test, which would be at the 165/166MHz crossover point.
