Brad-

I used AS 3 on my waterblock, about 3 weeks up and running now, I haven't experienced any bonus temperature drop after a period of running.

Temps are about the same as day one. I notice guys with air cool say the temp drops after about 50 hours of use.

Not here.
Jim

Pet,
Ok, maybe I'm just tired...but what is PIA???

And speak of the devil, Infonec left a message...the 8kha+ (#3) is ready to be picked up...so I'll get the fourth fan tomorrow after work.

Jim,
how does your Western Digital like OC'ing? (assuming you OC, I don't see it in your sig) My 10 gig Fujitsu died with a POP and puff of smoke last week. Big burnt "pimple" in the middle of one of the chips. (2 years old...never ran an UN-OC'ed day in it's life:D ) I bought the WD Caviar 40 gig, and I had to drop the fsb a couple mhz afterwards, just to keep it happy.

PIA= Pearson International Airport :D

Told you I was tired... lol

No, I go to the Markham store. I'm up on the east side of Lake Simcoe, just for reference.

oh kool :D

:cool:

I don't beleive that 50 hour thing either, it is feasible for a small decrease in temps, but nothing actually worthwhile

HeavyEquip - the Barracuda IVs will take all the FSB you can throw at them (200 verified). The Maxtor DX740s are known good into the 180 range. No good info yet on the new IBM 120GXPs.

<edit-dumb typo>

normally its the hdd controller that craps out before the hdd

Brad - notice in Heavy's case, he had to back off from a previously known good point with a new hard drive (i.e. his controller was verified to run at a higher rate than his drive would). In general, more recent HD controller ASICs will handle very high PCI bus speeds.

yeah, just use promise raid controllers, most are 66mhz pci

The fujitsu hummed along nicely at 195 mhz. Only problem I had with my system at that speed was using 1/5th PCI divider caused a HUGE memory bandwidth hit. I get much higher memory scores at 166x10.5x1/4 divider, than 194x9x1/5, and that's comparing same memory timing(SLOW). (I can't find my notes from the testing, but it's approximately 2200/1800 vs. 1900/1500 on the newest Sandra-buffered) At 166 I can get aggressive with this Micron, which increases my scores as well obviously(2482/2106), so the lower FSB works better for me anyway. I liked the bragging rights that 195x9, or 200x8.5 etc. gave me for awhile, but overall, my performance wasn't peaking.

Geez, long winded AGAIN. lol Yes, my HDD controller is reliable to 200mhz.

:D

Sorry for straying WAY off topic guys.

Oh yeah...at those really high speeds, my system ran 5C hotter too, so that's one more strike against 200 mhz in my book lol.

I don't know anyone else that does that around here :rolleyes: :p

If you were using the 1/5 PCI divider, then your IDE interface wasn't being pushed as hard as you are at 166 and 1/4 (40 vs 41.5). Won't the IWill allow you to use the 1/4 divider above 166?

And HTH is the PCI divider affecting your memory bandwidth? (especially since your core speed was about the same)

* wonders about the IWill board * :shrug:

My NIC and geforce2 MX can't handle anymore mhz,(41.5 and 83 mhz) so that's why I'm capped there, but yes, I can choose any divider at any speed, hardware permitting. Also my AXP is giving me all it's got at 1749.(at this temp anyway...I've gone faster on a cold night with the window open)

Ironic that I ended up back at 10.5 multi to get my ideal clock/fsb/pci, as that's the default.

As for the divider/memory bandwidth issue...it beats the hell outta me.
It is a known issue though,(Amdmb forums>Iwill) and it's rumoured to be resolved in a future bios. (Hmm, rumoured Epox 8KHA+ dividers too...this is me NOT holding my breath)

Also ironic, that I switched to the XP333 because of the divider, only to end up using 1/4.

It still baffles me though...CPU, DDR, AGP = Northbridge,
PCI, IDE, USB, Sound = Southbridge.
...how they are related via pci divider makes no obvious sense to me, but I speculate, (and I'm not in the mobo business, so I could be wrong) that the chipset communicates over the PCI bus, and some sort of bottleneck is occuring/getting more apparant with the bigger speed difference.

That hurt my head.

Just so your head hurts more >P
 

The NB and SB communicate over a special V-Link for the KT chip sets, not over the PCI bus (there isn't a PCI bus connection on the KT-266(A) NB).

Sandra's membench causes no HD activity or other NB<->SB communications to occur and any "background" activity wouldn't affect your membenches THAT much.

Sounds almost like IWill is either a) backing off the memory latency timings severely when you go to the 1/5 divider or b) they are setting the memory to run at the -33Mhz setting (the memory can run synchronously to the FSB frequency or asynchronously at +33 or -33Mhz to the FSB frequency) or c) both. From going back and relooking at the scores you posted, I would have to say either b or c.

If you know how to use Wcpredit, take a screen shot with the 1/5 divider turned on and the FSB at anything over 166Mhz and post it up and I'll be able to tell you what's really going on there ;)

As I'm sure you're aware, the ALi Magik is the underdog when compared clock for clock to a KT266A. Being the owner of two Iwill XP333's and one Epox 8KHA+, I can say that confidently.

The bios doesn't let you set much manually, CAS latency really, then there's 6 presets:Failsafe, slow, normal, fast, ultra, and ultra2

Cas 2 normal is the best I can get my 2.5 Micron to run at. It won't even post at 2.5 fast. While on the Epox, I can go Cas2, Turbo, 4way interleave...without using the 3/4 memory divider, so I know this memory is capable. Leaves me hoping that Iwill is figuring out a fix...the 1/5th divider performance is just a bios update away. lol

I don't have any experience with wpcredit, I found it a bit overwhelming actually. I had no idea what I was looking at, trying to do the HALT trick on the 266a. (I run a DC agent, so I don't idle anyway) I know I don't want to start playing with registers...too rich for my blood. One of these days, I'll unlock this processor again, I cleaned it off after I started running at 10.5 again, with plans to redo it later.( I couldn't use any higher multis)

Dayum, wasn't thinking and forgot the IWill used the ALI chipset. The ALI chipset DOES use the PCI bus to link the NB<=>SB. But there still should be no activity going on during membenches.

As far as Wcpredit, all you would have had to do is open it and do a screen capture of the window ;) But... in the case of the ALI chipset, I couldn't help :(

The ALI was originally designed for sdr + ddr ram, and it was the first ddr chipset available, When you remember that, it is actually really good

In Iwill's defence(and my own) the Revision C chipset, (this one) is 1000% better than the original. I waited for this board for a long time, based on the early buzz about it, and I was PISSED when they unveiled it...with the ALi. I went the KT266a route for awhile. A few surprising reviews later, I picked one up. Long story short...I own two now.

I'm kinda glad the KT333 wasn't as good as it should have been, self justification for my choice.:D

I have to admit though, it looks like I'll be jumping on the KT400(KT333a?)bandwagon next.:drool:

it's because the reviewers don't run 166/166, if they did there would be a huge increase in performance

heck or even 194/194 like i saw in one review... very respectable oc... have to have the chip unlocked tho... he got it to do 200/200 but the mem command time went up to 2 from 1 and seriously hurt mem performance... I'll probly go the kt333 route as the kt400 probly won't be out when i upgrade.


Cyph3r

Of course you do know that 200Mhz FSBs are being done on KT266A boards, right? (up to 205 range actually). At this point I really don't see what the KT333 will offer other than "official" support for 166Mhz. Until they get to the KT400, I'll personally stay with my hand (but then I usually only upgrade every other "generation").

As with the first round of TI4600 cards. I'll be patient and wait till they couple them with Samsung's newest 400Mhz graphics DDR memory (currently in the hands of developers for testing) :drool: :drool:

I have seen a few 8KHA+ WCPUid screenshots in the 180's, and 1 or 2 at 200...all I can do is think "lucky bastards".

My 3D card pukes after 166. (83) I notice these guys are running late model 3D. (Radeon 8500, Geforce4)

As much as my Geforce2 MX 400 is BEGGING to be replaced, I will never buy a "brand new, must have" card again. I bought my Geforce DDR(Annihilator Pro) a couple weeks after it came out, for just under $500 Canadian. Not 18 months later, the Geforce 3 was out, and I couldn't give my card away.

I've also heard a rumour that Amd is switching to a 166 fsb processor soon, (but NDA's are in place) in order to get a few more models out of the palomino, before thoroughbred is ready. (Tbred will be 166;) )

Kinda makes sense though, why would there be a wave of 333(+)chipsets starting? There's no way Amd would allow proprietary chip vendors to sell "overclocking" chipsets. All part of a grander scheme.

I'm sure glad the title of this thread "What do you think of this" sounds like an open discussion...I got in on rads, and now look:eek:

I think it is because amd wants to see just how much ddr2700 can be produced. if there is a shortage with it, right now you can just run at 133/133, big deal. While running at 166/133 with a TBred would just be really stupid, amd just wants to feel the market. Also, kt333, sis645, sis745, ali craptastic 1 will all allow amd to see how much pc2700 can be produced

:D LMAO

...hey, wait a minute.

;)

well back to MY topic :D

i got my HC and it needs some TLC

but now it looks like it is not copper :(

anyway...

what fans should i use?? i have 2 comair rotron muffin XL fans (120mm) which i will make shroud or use 3 ThermalTake 90mm or 6???

here's the heater core

Ok, seeing as I'm guilty for straying off topic, let me try to help here. Those 3 sections are roughly 4" squares. 4" = 101.6 mm.

I'd build a duct, and you can either taper it to fit 3 silent 80 mm's(my pick, no noise is good noise)...or flare it out for 3 120's. The 120's will cool better, but good luck on the noise. The 120mm Sunon I have (KD1212PTB1-6A) is loud (90cfm, 3100 rpm, 44.5 dba), and won't start at 7v.(Can "hot swap" it to 7v after it spins up, but just rocks back and forth if I start at 7v...and even then, it's not really quiet) The KD1212PTB3-6A is 69cfm,2400rpm,35.5 dba...but I find 20dba to be my high limit now since going to H2O. My 80's are in the 10-12dba range, and guestimating 20cfm.

I don't have any experience with 92mm fans, so I won't comment on them, but they are the closest to the 102mm x 102mm that you need for a perfect fit.

*EDIT*
It sounds like a good size too...you have roughly 308 square cm of surface, I have roughly 258. And I could fit your rad into my system way better than mine.:drool:

well the 120 arn't too very loud :cool: i have a 92mm that sux at blowing compaired to the comair's

I guess it boils down to personal taste. A noise-to-performance ratio.

All I know, my core does a decent job without the biggest fans...yours is going to kick ass with it's larger surface area, and some real airflow.

now i need to find some reostats that can handle 6 watt's :D