-
Re: Hardware pr0n...w00t!
Code:
read
CPU Motherboard Chipset Memory Read Speed
This Computer Gigabyte GA-K8NS nForce3-250 PC3300 DDR SDRAM 3090 MB/s
CPU Motherboard Chipset Memory Read Speed
This Computer Gigabyte GA-K8NS nForce3-250 PC3400 DDR SDRAM 3138 MB/s
write
CPU Motherboard Chipset Memory Write Speed
This Computer Gigabyte GA-K8NS nForce3-250 PC3300 DDR SDRAM 1217 MB/s
CPU Motherboard Chipset Memory Write Speed
This Computer Gigabyte GA-K8NS nForce3-250 PC3400 DDR SDRAM 1218 MB/s
latency
CPU Motherboard Chipset Memory CL-RCD-RP-RAS Latency
This Computer Gigabyte GA-K8NS nForce3-250 PC3300 DDR SDRAM 3-4-3-8 54.7 ns
CPU Motherboard Chipset Memory CL-RCD-RP-RAS Latency
This Computer Gigabyte GA-K8NS nForce3-250 PC3400 DDR SDRAM 3-4-3-8 58.4 ns
befores and afters respectively :unsure:
-
Re: Hardware pr0n...w00t!
Hmmm.
Not a real kick in the pants is it?
Write is slightly better and latency is worse.
I wonder if it's possible to tighten down the RAM timings and still boot.
Might have to add a tenth or so of voltage to the DIMMS.
-
Re: Hardware pr0n...w00t!
Quote:
Originally Posted by Peerzy
Even pushing it up a few hertz can cause massive ammounts of extra heat, thats how people fry parts, they go up from say 200 to 210 then 220 then 230 and everything is fine so they wack it up to 270 and parts get fried, be carfull, push them up at no more 10 on your clockspeed and use your BIOS to check temps, if they look good turn on Windows and fire up a graphically intense game (HL2 will do) if the game starts to look fucked with wierd shit happening then you have gone to far.
Not entirely true.
The heat generated by raising clock speed is roughly proportional the speed increase. But the rate of heat dissipation is proportional to the temp above ambient. The consequence is that a 25% increase in clock speed does NOT result in a 25% increase in temp above ambient for example.
That said, in order to achieve higher clock speeds it is usually necessary to increase the voltage. In this case the increase in heat generated is proportional to the square of the increase in the voltage. Add a clock speed increase too and that's when the damage is done.
@Clocker: Sorry, mine's a Via chipset. It wasn't stable @ 2.2GHz, but I don't know if that was the cpu or the memory running at 245MHz. I haven't seen any way of running the memory at a different speed yet. The temp was creeping higher than I wanted so I didn't bother checking anyway. I'm now back at 9x235 (or maybe 236, can't remember).
@Gepper, I assume you are running on a single memory stick, otherwise those timings suck.
-
Re: Hardware pr0n...w00t!
Quote:
Originally Posted by clocker
http://img.photobucket.com/albums/v7...ker/HWPorn.jpg
Well, it gets Sprocket all wet anyway....
New parts arrived yesterday, it was like Christmas in April except for the lack of decorations and insipid music.
Pictured above are the motherboard (Gigabyte GA-K8N Ultra-9), 4x512MB Patriot HP RAM(2,3,2,5 @1T), Athlon 64 3500+ CPU (90nm Winnie) and an Arctic Freezer 64 HSF.
Not shown are the XFX 6800GT (PCI-e) or the Plextor PX-17SA Dual layer DVD-RW.
Oh boy.
Tomorrow (Wed.) is my day off so I can begin the transformation.
First I have to strip out the Chaintech 754 setup from the Silverstone case and build it into a new box- my brother has bought Sprocket's guts for his boys new PC.
I hope to have all that up and running by tomorrow night.
Then I can begin to reassemble my baby in her new glory.
The first build will be pretty quick and dirty cause I have some more changes in mind- the new Antec P-180 case looks very interesting and I'm teetering on the edge of watercooling again, so I can't (won't) really tart up the wiring till I decide what to do ( the Antec has a completely different layout than my current TJ-06).
The watercooling is more for noise prevention ( which is the entire point of the Antec case) than performance
(see no point in OCing this setup really), so the Freezer 64 is just a (temporary) experiment.
Updates will follow as time allows.
How you have changed in such a short time... :no:
-
Re: Hardware pr0n...w00t!
-
Re: Hardware pr0n...w00t!
Quote:
Originally Posted by Virtualbody1234
How you have changed in such a short time... :no:
Forgive me VB, for I have sinned.
Anyway, what's a few hertz among friends?
It's hardly as though I've taken this PC by the neck and strangled every last cycle from it's body now, is it?
All I've really done is tighten her up a bit...a "tune-up" if you will.
A theoretical question...
If stock speed is 2.2GHz, achieved with a 200x11 combination, is 220x10 considered overclocked?
If Sprocket crashes out of Prime and I'm not there to hear it, is there any sound at all?
Why is there air?
-
Re: Hardware pr0n...w00t!
Quote:
Originally Posted by clocker
A theoretical question...
If stock speed is 2.2GHz, achieved with a 200x11 combination, is 220x10 considered overclocked?
If Sprocket crashes out of Prime and I'm not there to hear it, is there any sound at all?
Why is there air?
Yes. The 220 overclocks some parts of the system.
Yes. It makes a crashing sound.
For air cooling of course.
-
Re: Hardware pr0n...w00t!
i turned my RAM voltage up yesterday and got no improvement, i decided i wouldn't bother changing it until next time i restart.
this morning i got up to find my pc reset itself. i hope it's not dead :fear:
-
Re: Hardware pr0n...w00t!
Minor tweakage update.
_I have yet to discover any set of circumstances that will allow POST using 1T memory command.
_nForce 4 chipsets don't seem to respond to higher tRAS settings as the earlier 2/3 versions did. Changing from 5 incrementally up to 11 gave no significant reproducable gains.
_Upping the RAM voltage by .1v did result in a consistent latency drop ( now under 46ns), but this may only hold true for my combination.
Within the next day or two I'll be able to test 2x1GB sticks in place of my 4x512MB (same Patriot EP modules) just to see what happens. If nothing else, this would free up some breathing room around the DIMMs, possibly resulting in lower RAM temps (although I have no idea what the current temps are and have no way to monitor them).
We have had severe compatability issues with 1GB sticks on nForce 2 & 3 boards so this a more out of curiousity than anything else....I don't expect any significant change in performance.
I have a new build that the customer had specced for 2GB of RAM. I installed 4x512MB of the same Patriot that I just bought.
He was horrified when he saw the machine...
"Now I can't upgrade to four GB if I want to!"
"Why do you need more than 2GB?"
"I don't know, but I want the option".
:frusty:
So I'm bringing in two 1GB sticks, which is OK since I'll get to try them out first.
All of our 939 builds have had a minimum of 1GB RAM and the last three (mine included) have run 2GB.
The 90nm Winchesters installed on a nForce4 board with 1GB+ of RAM is a wonderful combination.
It'll crunch through your email in no time flat..... :P
-
Re: Hardware pr0n...w00t!
Well, I've been a bad, bad boy.
Flogging away at the ole PC...made some progress too, it seems.
Many, if not all, of the preconceptions that I was dragging into this from my socketA days were flat out wrong.
Anyway, here's some results....
http://img.photobucket.com/albums/v78/clocker/test1.jpg
This is a shot of my entry into the Bit-Tech SuperPi Challenge.
Near as I can see I should be in either 3rd or 4th place overall (1st place for an air cooled rig).
Here are the Everest memory tests...
http://img.photobucket.com/albums/v7...er/memread.jpg
http://img.photobucket.com/albums/v7...r/memwrite.jpg
http://img.photobucket.com/albums/v7...er/latency.jpg
All this (16% increase in clockspeed, 11% decrease in latency, 24.4% increase in write and 20.3% better read) and the temps have hardly budged...
http://img.photobucket.com/albums/v7...ker/temppc.jpg
Can't say I trust this monitor too much, but it's consistent with BIOs for whatever that's worth.
Just for grins, here's the Systems Comparison from PCpitstop.
http://img.photobucket.com/albums/v78/clocker/pcps.jpg
They are having a TopDog Competition in which 1st place wins a big Raptor.
Me wants it, but I have a ways to go to be in the hunt.
Then again, I haven't even begun on the video card...there is a lot of room to bump that score.
-
Re: Hardware pr0n...w00t!
Clocker, I keep hearing you post that you don't believe the low temps.
Do a simple test... Feel the heatpipes at the base of your CPU cooler with your finger.
Tell us what you notice. :)
-
Re: Hardware pr0n...w00t!
Erm....nothing.
My RAM feels hotter.
-
Re: Hardware pr0n...w00t!
So what have you changed to achieve this big jump in cpu speed?
-
Re: Hardware pr0n...w00t!
Let's see if I can remember....
RAM:CAS: 2.5,3,3,8
1x HT
2T enabled
Clock max 200MHz
CPU:stock voltage
+.1v RAM
10x257
Doesn't really seem like much, but it's taken forever to get here.
Although this setup will run SuperPi no prob, I suspect that it wouldn't Prime worth a damn.
Let's find out....
-
Re: Hardware pr0n...w00t!
Isn't 1xHT rather slow? I thought the default was 5x (@200MHz=1000MHz) so wouldn't the nearest equivalent be 4x (@257MHz=1028MHz).
Or is it the ratio of HT speed to clock speed, with the 5x multiplier still in there effectively giving you 1285MHz?
-
Re: Hardware pr0n...w00t!
Quote:
Originally Posted by lynx
Isn't 1xHT rather slow? I thought the default was 5x (@200MHz=1000MHz) so wouldn't the nearest equivalent be 4x (@257MHz=1028MHz).
Or is it the ratio of HT speed to clock speed, with the 5x multiplier still in there effectively giving you 1285MHz?
You got me there, Lynx.
All I know is that lowering that setting made it possible to start upping the FSB.
I ran the Everest memory benchmark every step of the way, not being willing to sacrifice RAM performance for sheer clock speed, and the results just kept improving till I got to the point I showed you.
My limit seems to be 260 FSB now.
I have tinkered a bit with various combos of multiplier/FSB but 2.6GHz is the wall so far.
My board does not offer the myriad voltage settings found on the DFI LanParty series and that could be the problem, but I'd prefer not to crank up the heat anyway so the point is moot.
My available voltage settings are suspect anyway (just like my temps!)...changing the vCore from 1.3v to 1.5v doesn't alter the reported voltage in either Speedfan or Everest- it is steady at 1.39v.
Could this be a faux setting, merely eye candy or a function of the chip itself?
My highly sensitive finger ( recommended by VB) seems to corroborate the temps I'm reading, but I don't see how a jump from 2.2 to 2.57GHz wouldn't result in a bigger jump than 2C, but as I said, my preconceptions are all wrong apparently.
As I suspected, Prime95 crashed out within 3 minutes at my highest clockspeed...I was stable enough to run SuperPi, but that's about all.
I've ratcheted down to 2.4 GHz and have been stable for over an hour so that might be the final resting place.
We'll see.
Edit: I'm not sure what difference (if any) it makes, but I have also cut my running services down to the bone for the benchmark testing. I'm working on a gutted XP install disk using nLite just to see if that helps too.
http://img.photobucket.com/albums/v7...askmanager.jpg
-
Re: Hardware pr0n...w00t!
Just ordered nearly $400 worth of watercooling gear.
WTF am I thinking?
My CPU really doesn't need it and I expect little drop in running temps.
My GPU on the other hand, should benefit mightily from a drop in temps.
It currently runs in the mid-50's, peaking at 59c after a round of benchmarks (Aquamark 01 & 03 and 3D mark 05). This is with the Zalman copper flower installed.
I'm reading of max temps dropping into the 30's using my chosen DD GPU block.
We'll see.
I'll be using a BIX radiator which may run passively depending on how the results go.
I'll keep ya'll posted.
Finally broke 76K in Aquamark03 this morning....w00t!
Hope to hit 80K on water, but that may be a pipedream.
-
Re: Hardware pr0n...w00t!
was the pun intended :ermm:
well it'll look uber cool and it'll be quieter because you can slow/disconnect some fans. i think :unsure:
-
Re: Hardware pr0n...w00t!
Quote:
Originally Posted by clocker
... WTF am I thinking? ...
I don't know. Maybe you could inform us?
-
Re: Hardware pr0n...w00t!
Quote:
Originally Posted by Virtualbody1234
I don't know. Maybe you could inform us?
When I figure it out I'll let you know. :blink:
-
Re: Hardware pr0n...w00t!
Quote:
Originally Posted by clocker
When I figure it out I'll let you know. :blink:
:lol:
-
Re: Hardware pr0n...w00t!
I am fascinated by hardware and pictures don't satisfy the craving...so I am compelled to actually buy this crap.
This thread isn't named "Hardware pr0n" for nothing, you know.
It's a sickness.
That's why I stand on streetcorners holding a sign that says:
"Will Work For Raptors". :P
-
Re: Hardware pr0n...w00t!
Quote:
Originally Posted by GepperRankins
Code:
read
CPU Motherboard Chipset Memory Read Speed
This Computer Gigabyte GA-K8NS nForce3-250 PC3300 DDR SDRAM 3090 MB/s
CPU Motherboard Chipset Memory Read Speed
This Computer Gigabyte GA-K8NS nForce3-250 PC3400 DDR SDRAM 3138 MB/s
write
CPU Motherboard Chipset Memory Write Speed
This Computer Gigabyte GA-K8NS nForce3-250 PC3300 DDR SDRAM 1217 MB/s
CPU Motherboard Chipset Memory Write Speed
This Computer Gigabyte GA-K8NS nForce3-250 PC3400 DDR SDRAM 1218 MB/s
latency
CPU Motherboard Chipset Memory CL-RCD-RP-RAS Latency
This Computer Gigabyte GA-K8NS nForce3-250 PC3300 DDR SDRAM 3-4-3-8 54.7 ns
CPU Motherboard Chipset Memory CL-RCD-RP-RAS Latency
This Computer Gigabyte GA-K8NS nForce3-250 PC3400 DDR SDRAM 3-4-3-8 58.4 ns
befores and afters respectively :unsure:
ive got a stick of that £25 ram and now:
read - 3005mbps
write - 1185mbps
latency -60ns
it wouldn't work in the second slot strangely, my pc kept resetting when booting to windows.
so now i have
512mb pc3200 400mhz ddr
empty
512mb pc3200 400mhz ddr
does anyone know if i can get better performance or is it just one of those things?
-
Re: Hardware pr0n...w00t!
What program is generating those results?
-
Re: Hardware pr0n...w00t!
everest home edition :unsure:
-
Re: Hardware pr0n...w00t!
Good.
The PC I just built for my brother is also a nForce3-250.
Tonight when I get home I'll install Everest on it and see what kind of results it gets.
-
Re: Hardware pr0n...w00t!
Dave, what's your clock speed now? Has it reverted back to 200MHz?
I hadn't realised that the K8NS was a S754 board, when I looked it up I got the impression it was S939, but I realise now that I was looking at the K8NS-939 (damn stupid way to number boards :frusty: ) so you aren't going to get the high dual-channel speeds.
I think if you check back I said that the cheap mem was ok if you didn't plan on OCing, but you should still get an advantage if you ditch the page file.
-
Re: Hardware pr0n...w00t!
Boy o boy, nice kit, that will shift a bit :cool2: :yikes:
kabar
-
Re: Hardware pr0n...w00t!
Quote:
Originally Posted by lynx
Dave, what's your clock speed now? Has it reverted back to 200MHz?
I hadn't realised that the K8NS was a S754 board, when I looked it up I got the impression it was S939, but I realise now that I was looking at the K8NS-939 (damn stupid way to number boards :frusty: ) so you aren't going to get the high dual-channel speeds.
I think if you check back I said that the cheap mem was ok if you didn't plan on OCing, but you should still get an advantage if you ditch the page file.
it's at 205mhz
i haven't noticed any downside at all, i just want it all :lookaround:
in fact it's fantastical. now games don't jerk when theres a new sound effect or owt and quitting them gets me straight to desktop or IE without me having to spin the hard-drive :01:
-
Re: Hardware pr0n...w00t!
OK Gep...
With a 3000+ Clawhammer running on a Chaintech ZNF3-250 ( stock settings):
read:3062
write:1225
latency:51.6
Ram is Mushkin Blue Line PC3200 (1GB, located in slots 1 & 3).
Not really that much different than your results, so I guess we are both in the ballpark.
BTW...
Started getting the H2O stuff in today.
From FrozenCPU I got the 3/8"id Tygon tubing and clamps.
Unfortunately, the PolarFlo CPU block and pump ship direct from PolarFlo and I have no tracking # so no idea when they might show up.
The GPU block from DangerDen also arrived, and I must say it's a very impressive bit 'o kit.
Now for a wee hypothetical question...
By far the hottest part of the PC is the video card.
By comparison, the CPU is absurdly cool.
I'm wondering if it might not be feasible to run the loop thusly- pump>radiator>vid card>CPU>pump.
This would (presumably) allow for maximum temp reduction of the GPU ( which I'd like) and maybe give up a degree or two on the CPU (which I can afford).
Then again, maybe not.
I know the only way to really find out is to try both ways, but setting up a water loop and then breaking it down is quite the PITA and sheer inertia means that the first arrangement is prolly the one I'll stay with, so I'd like to get it right the first time if possible.
The floor is open...
-
Re: Hardware pr0n...w00t!
Spoiler:
Show
With a couple of Y connectors you could possibly cool both in parallel.
:shifty:
-
Re: Hardware pr0n...w00t!
You could also leave out the CPU from the loop.
-
Re: Hardware pr0n...w00t!
Quote:
Originally Posted by lynx
Spoiler:
Show
With a couple of Y connectors you could possibly cool both in parallel.
:shifty:
Yes, but at the cost of more complicated tube routing.
One of the big advantages of the components I chose is the barb arrangement and the resulting short hose runs.
Quote:
Originally Posted by Virtualbody1234
You could also leave out the CPU from the loop.
I thought of that option too, but would like to see if I can eliminate as many fans as possible.
-
Re: Hardware pr0n...w00t!
Parts is parts is parts, etc.....
http://img.photobucket.com/albums/v78/clocker/gear.jpg
Finally have all the goodies in my possession, ah! I love the smell of new hardware in the morning!
The basic layout will be easy, but there are a number of custom fittings that I'll have to machine so it might take a day or two to get to the test phase.
The NV-68 video block is absolutely massive and will require some sort of support bracket for the card...I have a plan, we'll see how it works.
I want the tube runs to be as short and elegant as possible (hence the choice of 3/8" tubing ) and the only real unknown at this point is the fill/bleed port.
I have two options available and I won't be able to decide till I start to mock the components in place (the first config would be ubercool, but may prove awkward to use. The second would be more traditional, but probably more practical.)
After all the necessary drilling/fitting is done the whole case must be stripped for powdercoating.
This is actually more involved than it seems since at the same time I'll be replacing the drive bay cage with a different arrangement (I absolutely HATE "tool free" mounting....).
Lastly will be the wiring. I've decided to commit these components to each other and the case for good, so I can go ahead and cut/refit the wiring harness to exact lengths.
This will be fun. :D
-
Re: Hardware pr0n...w00t!
All major components mounted, but I can't bleed the system properly.
I hate this crap.
Got parts this AM to cobble up an air trap.
I can already tell that the DD GPU block is going to be a bear to bleed...there is a real air trap around the processor finning and my most extreme contortions only managed to partially dislodge the bubbles.
With luck the trap will solve this problem.
Later.
-
Re: Hardware pr0n...w00t!
-
Re: Hardware pr0n...w00t!
holy crap clocker.... i have the exact same speakers...
-
Re: Hardware pr0n...w00t!
Quote:
Originally Posted by Monkeee
holy crap clocker.... i have the exact same speakers...
You wanted better cooling for your vga card, why not follow clocker's example on that too.
Monkeee see, Monkeee do. :blink:
-
Re: Hardware pr0n...w00t!
Quote:
Originally Posted by Monkeee
holy crap clocker.... i have the exact same speakers...
Ah, but mine are watercooled.
I can play Bohemian Rhapsody at full volume with no artifacts.
-
Re: Hardware pr0n...w00t!
Quote:
Originally Posted by clocker
Ah, but mine are watercooled.
I can play Bohemian Rhapsody at full volume with no artifacts.
At full volume. I'm sorry I just don't belive it.
