Sign in with
Sign up | Sign in
Your question

hit a brick wall

Last response: in Overclocking
Share
February 15, 2006 7:15:26 AM

hi guys how you all doing

I am busy overclooking my rig from a X2 3800+ to 5000+ (2000MHz to 2600MHz) and that is where it stops. It is stable at this speed...

As soon as I go to 261MHz it doesnt even boot

I have a feeling that it is ram... I have upped my VCORE and still nothing
I have also dropped my HTT link and still... you guessed it... nothing

I have seen AMD's on Futermark forums clocked at speeds of up to 2998MHz most are running at about 2800MHz

I am looking at either Corsair XMS TWINX1024-4400C25PT or
Kingston HyperX KHX4300K2/1G

My Rig

AMD X2 3800+
ASUS A8N32-SLI Deluxe
256 x 4 ddr 400
ASUS N7800GT extreme
X-Fi Platinium
Antec P180 Advanced Super Mid Tower
Antec 500W smart power 2

Will my mother board support this ram on ASUS web site it says that it is Corsair XMS-3500LL PRO Ready

More about : hit brick wall

February 15, 2006 7:43:22 AM

If your says it support the Corsair XMS pro series then chances are it does and they have tested it. The XMS pro series are very good at overclocking. You should get 2Gb to get a significant performance increase over your current set up.
February 15, 2006 8:47:23 AM

Why do people think Asus = Good Overclock ?
Asus = Good for beginners overclocking perhaps.

Try turning all their 'advanced' [cough] features.
- AI NOS (Off)
- PEG Link Mode (set to Slow in BIOS)
- etc

Try cooling the chipset better.

Better RAM *might* help, but don't count on it.

You'll want an Abit or DFI board most likely, pref a pretested / tested at overclock already one.
Related resources
February 15, 2006 10:12:19 AM

Getting good ram like corsair is good for adjustment of timing, but good cooling is very critical for cooling the CPU GPU and lets not forget the other components that this power is created from, the coils, transformers, and voltages regulators a good air flow will allow proper cooling of those componts. Higher frequencies produce more heat and more stress on the generation station that produces those frequencies and when these components are ventilated correctly the better LOL you will receive when over clocking.
I would like to point out to people who might not be aware that when they get their new processor and boards vcards it is important to know what all factors of the board are, specially the timings for CPU, RAM, HTT, Duel channel, test what timing work well with the video card and the other components like CPU, RAM, HTT monitor the temps in the testing phase.
I say keep the parts in the mid 40 to low any higher then these temps will cause weakend parts specially the coils, transformers.

Everytime you make changes to the frequencies timings. Benchmark with this principal you will always be successfull in overclocking. This is over 20 experience giving you informantion.
February 16, 2006 7:23:55 AM

Quote:
Higher frequencies produce more heat and more stress on the generation station that produces those frequencies and when these components are ventilated correctly the better LOL you will receive when over clocking.....

.....specially the timings for CPU, RAM, HTT, Duel channel, test what timing work well with the video card and the other components like CPU, RAM, HTT monitor the temps in the testing phase.
I say keep the parts in the mid 40 to low any higher then these temps will cause weakend parts specially the coils, transformers.

Everytime you make changes to the frequencies timings. Benchmark with this principal you will always be successfull in overclocking. This is over 20 experience giving you informantion.


Sorry 'gomerpile' but your above post is going to confuse REAL PEOPLE one day:

- By the looks of it you hadn't seen a processor 20 years ago, let alone overclocked one 20 years ago. In 1986 do you even know which processors were available on the market ?, How much they cost ?, Stock availability ?.... Heck mentioning using benchmarking 'trial & error' runs for [cough] stability testing implies you haven't been doing this very long at all. (Trial & Error is the least time effective way to confirm stability).

- Boards don't have timings as such, only RAM really does, generally speaking Video Card RAM timings do not need to be changed for vid-card overclocks either, (But this can be done using ATI Tray Tools anyway to raise VidRAM clocks even higher, just like mainboard RAM). Mainboards have mostly multiplier options and clock speed adjustments.

- It is spelt 'Dual Channel', anyone that is over 20 years old (you'd need to be 34 or so to have 20 years experience in IT) would clearly know this and never typo it. Only a 9-14yo would typo it - (people here tend to pay attn to this in particular)

- Temps below 40 C while overclocking on air, yeah good one.... I suggest anything under 70 C is stable, ideally 65 C.... however up to 85 C may be safe in some circumstances. (eg: typical GPU runs at 80 C at factory settings). Athough 85 C is the orange/red zone. Hell, why not recommend sub zero temps for normal operation while you're at it ?, while pretending condensation is a null issue.

- Most parts are designed to be run in an ambient case temperature of 55 or 60 C , and if that is the ambient temprature, then anything generating heat may well get much hotter. IN LAYMENS TERMS PC PARTS ARE NOT HUMAN BODY PARTS, 40 C is 'BARELY WARM' - Do people actually believe this **** still ?

- What is this "generation station" you speak of ?, No please don't answer that I've read more than enough tripe already.

- "components are ventilated correctly the better LOL you will receive when over clocking" - because I always want 'better LOL' when overclocking.

- Can anyone pick anymore faults with some posts around here ?, Should TomsHardware give out prizes for removing these posts ?, Should posts like the above be able to generate a negative post count / limit / auto-ban on users, while having a 2 week 'no post' lock for new users (to stop foolishness) ?
February 16, 2006 8:27:30 AM

hehe tabris :lol: 

gomerpile, sounds like your LOL is overheating the flux capacitor in your BIOS matrix which is sending negative energy through your generation station. Try pouring custard over your monitor... 8)

Jayman, any more specifics on 'doesn't even boot'? Only ask because I had a similar problem with my ASUS A8N-SLI Deluxe and it was the northbridge overheating. I think you get the improved heatpipe design on the A8N32 but could still be worth looking at.
February 16, 2006 9:35:50 AM

I just (like 30 min ago) got my Opteron 270's to 2.41 GHz each (just above Opteron 280 speeds).

Aggregate gain was +1.64 GHz in processing clock speed (over the 4 cores).

Wasn't going to, but nVidia nTune made it too damn easy, and used to overclock heaps (eg: +50% on Duron 600, minor tweaks to get there only with cheap RAM) when I had less money.

Ahhh, bleeding edge performance (Opteron 280 level), Not the insane destructive edge performance that needs liquid carbon cooling, --- It has been too long.
February 16, 2006 10:34:22 AM

Old school "A" lets see you top this dude







the specks, 255MHZ, 11.5, 2volts, on a xp3200
Vcard timing tRc=17,tRFc=19, tRAs=12,tRP=9,tRCDRD=9,tRCDWR=9,tRRD=6
volts 1.75 at 485 mhz ram 1300 mhz
2 enermax 600watt psu
2gig corsair twinmatched pairs
Runing temps full load is in the pic

second system is tyan tomcat and still buying parts, when fully tested I will give the specks currently I am building the case now,
current the xp system is the best bench summited as of this date for an xp3200 system even the 64bits in the 3400 range dont even come close.
Timing mean lots if one knows what they are doing its the timing that give the the extra boost people need to really oc. Without the faster timing you can only go so far than the computer stops, data error and what ever else you can think of but with the know how you can modify the bios for the system and vcard, you can change the timing so the speed and power is in time when called upon and without it you will only go so far and than you will get what jay gets. I have an A++ an MCSD behind all this.
February 16, 2006 10:54:14 AM

Quote:

Old school "A" lets see you top this dude


Fair enough:

JPEG: (275 KB)


And a link to the full quality PNG: (387 KB)
http://users.on.net/~darkpeace/forum_images/hit-a-brick...

Need any more proof ?, because I can take screenshots all night if need be, but we are both hard-core and/or quad-core, etc, etc

NUMA also gets me just shy of 10 GB/sec sustainable memory throughput. (Each processor has its own memory, and can aggregate performance using the opposite processor like a 'northbridge' to the 2nd memory node).

You do have SiSoft SANDRA right ?:
- http://www.sisoftware.co.uk/
- Mirrors: http://www.sisoftware.co.uk/index.html?dir=dload&locati...

EDIT: I wasn't aware the A++ (Hardware I assume) and MCSD had specialist sections in x86 chipset hardware intricacies, including the various timings, etc. :p  . Still 255 MHz (base, so 510 DDR FSB right ?) on an Athlon XP is quite impressive... sure it still limits it to about 4 GB/sec to/from northbridge, and shares said 'bandwidth' with RAM I/O, AGP I/O, and other requests (to southbridge, IDE / RAID controller, etc) as the northbridge becomes a bottleneck (at times anyway).

Figure it is cheaper to move to x64 (AMD64 / EM64T), eg: an Athlon 64 3500+ would give similar performance to your rig, at a lower TCO-$. (But they may not have existed when you build the Socket A system above). Without the risks of 'extreme' cooling measures, etc... still a cool system.

Sorry, your (far above) post just seamed really weirdly worded for someone with technical skills (displayed above).
February 16, 2006 11:07:57 AM

Well in a few days my tyan will be done and that will. I'm sure I will match the speed but I dont see any customs to this other than MHZ, if you were to change the timing you would get more sure the xp3200 wont match the speed.
In terms though I dont see your system ocing anybetter in terms of what you have and what I have. MY point is I"m far from old school and Jay needs to adjust his timing to suit his ocing ability for max
February 16, 2006 11:15:17 AM

Changing the timings of Registered RAM (ECC or not) is not usually a good idea. Registered RAM adds a 'transparent' CAS tick anyway, the upside it can move to 8-16 GB (Well up to 32 GB as high density Registered modules were available up to 4 GB in size at time of purchase, but they had lower timings and usually lower clock speeds, PC2100 - PC2700).

Not that RAM performance is weak though, as I've already reduced some of the memory controllers (int in CPU die) timings from the defaults, but not the timings on the RAM itself, as it can sustain 10 GB/sec (SANDRA, but peak is still 12.8 GB/sec), only 80% efficient true, but most memory reads are 64 - 512 bytes which is still 95% - 99.375% efficient (compared to peak ) with the stock timings anyway.

The only Video card RAM timings raised are: TRP (Timing Row Precharge), from 5 tp 6, and TRAS (Timing Active to Precharge), from 6 to 7, as this enabled slightly higher overclocking for the video memory and increased stability from the factory settings too ;)  . Have actually lowered TW2R (Write to Read turnaround) from 7 to 1 (within the same bank of video ram only though). Still only run the card at 400/1000 though, but it is only a Radeon X800 XL (400/980) anyway. :p 

Replacing it with a GeForce 7800 GT around March most likely. (after CeBIT) - Don't game that much.

Minor edits, added TRP, TRAS before & after. Changed video-ram clock speeds to their DDR figures
February 16, 2006 11:19:49 AM

yes if you can run at the lower and maintain the higher MHZ than you are doing well but in my case I had to go with the slower timings. To produce the MHZ. Second is I use MR.Bios editor to modiy the system performance all in all this aint old school stuff dude.
!