HP 8510P new Bios !!! improves HD 2600

i use avg , spybot and spyware blaster , with the last version of sygate, light on resources and lets me make rules i want . firewalls really age well and sygate was great .i find norton interferes with to many things

 

I ran memtest86 yesterday and it showed me a ton of errors. My RAM is broken I guess...

 

I´m a little confused right now. Did some testing with several programs to check the temps of my gpu.

as i´ve reported earlier i watched the temps with the amd gpu clock tool. This gave me rather good ( too good to be true ) results. Even with the highest possible clocks for my hd2600 ( 636mhz/684mhz), the temps of my gpu never went above 35°C.

did some testing today with speedfan 4.33 and everest ultimate 4.20.227.0.
Everest did show about the same temps for my gpu ( around 33°C).
When i look at Speedfan and watch the temp2 thats supposed to show the gpu temp, the temp is around 65°C. Twice as much as the temps shown by everest and amd gpu clock.

i guess something must be wrong here. either my gpu asic is broke or what i suppose the temp2 of speedfans monitoring tool is not showing the GPU temp, but the CPU core temps.

i´ve made a screenshot for comparison between everest and speedfan.

right now i can only guess whats wrong here. If speedfan does report the gpu temp it´s probably temp4, this comes at least close to the results i´ve had with everest and amd gpu clock tool.

If this is correct my gpu temp is actually around 30°C - 35°C when it´s overclocked to the maximum my hd2600 can go.

i hope someone with more knowledge can shed some light on this.





@ vpicbim

sorry to hear your ram is not the best. is it the original one from hp ?
If it´s the original ram you´re having problems with, maybe you can get new ram from hp.
but you will not have any guarantee if it´ll be better. i remember the memory in my hp was from hynix. don´t know if this is any good, but since ram is absolutly cheap right now, i´ve bought 4 gig ocz about two weeks ago. and i´m not having any problems with it so far.

 

@Schoko

yes, it's the ram that came with the computer. I'll call customer service after the holidays and see if they can replace it.

I was thinking of buying 2x 2GB but will WinXP Pro 32bit even support it?
Is it a problem having ram sticks from two different manufacturers?

 

Xp pro will only support around 3 GB Ram. But thats more than enough for XP.

See it as an investment for the future. I purchased the 8510p wih vista business 64. for now i rather work with xp, since i earn money with my computer. But when vista 64 is a worthy upgrade, and my apps are working "problem-free" i´ll switch.

I bought the 4 GB for Vista, right now i don´t need them, but as soon as i switch , it´s good to have them , since i´m working with some highend 3d apps, and do a lot of rendering and photoshop work.

about using ram modules from different manufacturers, it´s no problem. the only thing you need to watch out for is buying sodimms with 667 mhz. you can even buy the 800mhz sodimms. but it´s not worth the extra money to buy the 800mhz sodimms, the santa rosa plattform will only clock them to 667 mhz.

 

my gpu runs much hotter more like 60 -70 when taxed 35 when idle

 

strange thing...i can run 3DMark with oc'd to 600/700 but if i play Call of Duty 4 at 600/700, 575/675, 550/650 it keeps crashing after a few mins
So i think my card isn't that overclockable after all...

 

I just clocked my card at around 630/740 and played the Unreal Tournament 3 demo for a couple of hours with no problems at all.

Maybe there's something wrong with CoD4?

What temperatures does your card show?

@Schoko

I was kinda wondering why you have 4GB ram but run winxp 32bit. I guess that explains it My next investment is probably gonna be a faster HDD. 5400rpm is quite a bottleneck.

 

@ vpicbim
good to see you have these high clocks applied and no crashes.
i am wondering whats the problem with my card. whenever i clock the core higher than 640 ( mem above 680) my system freezes or crashes when starting a 3d app.

i thought i might have just hit the voltage limit there, but maybe it´s just that the core of my hd 2600 won´t allow higher clocks.
maybe i´ve just got a bad card. But i´ve got to admit it´s not as bad as i describe it. It´s just that i can´t clock it higher. But when i´m below the 640 for the core i can play crysis with the higher clocks and my laptop is running stable as a rock.

But still i´m no sure about the temp readings i get. It rather disturbs me when people tell they have around 60°C, and my card is just showing around 35°C.
i hope it´s not the thermal sensor thats broke.

 

nope, nothing wrong with CoD4 since i can play it on standard clocks for hours. Still gonna try it with other games thought..

 

I tested it again. I set the clock to 634/738 and played UT3 and Need for Speed Pro Street for about an hour with no crashes or artifacts.

Everest showed a temperature of 31*C. I'm not using a cooling pad, but I put a book underneath the laptop to increase airflow.

I'm sure I can push the system even a little bit further


/hmm... those readings are almost too good to be true. According to Everest, the idle temperature is at 23*C which is slightly above room temperature. CPU shows 49* and motherboard 46*. It's hard to believe that the GPU is that much cooler, so there must be something wrong with sensor I guess.

 

that´s exactly what my temp readings with everest and amd gpu tool show

~23°C when in idle mode
~33°C - 35°C under load (specviewperf 10, maya, crysis, 3dmark)

i´ll search for some other programs to measure the temps. though these temps readings are looking good ( too good to be true), i don´t believe the gpu is actually this "cold" .

would be nice to have some more temp readings from others here to compare.

 

Yeah that's really weird, especially after seeing those screenshots schoko. I honestly have no clue

 

schoko: Is your room cold due to winter? I know my room is quite cool so it brings down my computer temps.

 

yeah, well cold is not the actual term i would use to describe it, since i like it generally a little warmer. But i´m not living in a freezer.

i´ve got around 20-22°C, so that can´t be the problem. have you guys measured your card´s temps with everest and amd gpu clock tool too ?
if so what are your temp readings ?

// edit - the screenshots are actually only one, made at the same time, cropped it, so it would fit this board preview //

 

I'm getting 2700 in 3dmark06 for some reason, previously i had 3300. I flashed the bios, installed 7.12's and now im getting 2700, any ideas why?

 

Just wanted to say this is an excellent thread, the scores you guys are getting are phenomenal!

Your temperatures are really good as well (looks like ~70-80 from what i've seen). This is quite good compared to the 8600GT which hovers at around ~90 degrees celsius under load. 65nm cards FTW.

 

This is exactly what Ive noticed when discussing with Aceron. Even if Aceron was right, its very simple to test it. Just touch the area just underneath HP sign between presentation and mute buttons where the GPU sits so U will see that the GPU temp cant be 60C!

Anyway, lovely scores, cant wait to o/c mine BTW, what sort of improvement did U see schoko when playing Crysis with stock and o/c clocks??

 

2700 wow that is a big difference. i suggest you go back to original bios unless you were having those vpu crashes, because you dont really need to update bios unless there is a problem that needs fixing.

as for me ive beeen getting the same score with and without bios, and with and without the new ati drivers. my score has been hovering around 3040. ive tried doing a fresh install and still the same results.

other than that im quite happy because i have no more crashes due to the new bios.

any ideas with the low marks were getting?

 

eeh i believe you always have some bios you can't work on a vid card today with no hardware controlling basic program unless you're doing everyting in bytecodes and asm

concerning the question - have you checked how many processes there are and memory your os take? there might be problems with less mem available to the testing utility.

 

Maybe your settings for the drivers are default set to be optimal settings not best performance or sthing like that??

 

Yeah believe me I'd rather the temps be lower than what I'm seeing, haha. Regardless of whether it's 30 or 60 though, still much cooler than the 8600 GT GDDR3

bharris - I would confirm you're on the latest bios, then user drivercleaner and wipe your GPU drivers. Then apply the latest ATI Cats (7.12?) with the modtool. I'd then run 3dmark06 @ 1280x1024 in high performance mode, also running gpu-z or ati tool first to confirm you're flying @ 500/600 mhz.

If you still get weird scores, perhaps contact HP :\

 

hey guys what is your virtual memory set to with those who have 3GB of ram? is it the recommended number?

 

Highest possible

 

When you ask for the a reasonable setting for your swapfile the answer will be around 2 up to 2.5 times the amount of ram you have.
But this is an old rule, since we have a lot more ram than like 5 years ago.
It was common to have a swapfile of around 2048 - 3072 MB when having one GB of ram.
This would garantee that you can work with big files. But since we are already having between 2 and 4 GB of Ram the old rule does not make that much sense anymore.

one example : you have 3 GB of Ram , now make your swapfile twice as big, and you´ll end up with 9 GB ( 3 GB Ram + &GB Swapfile). I guarantee you , as long as you are not making extreme calculations like predicting the wheater for the next 3 weeks or rendering an image with a size of 16000 x 12000 pixel you will not use a lot of your swapfile.

just can speak for me, but i´ve got 4 GB Ram, Windows XP 32 does recognize around 3 GB. Since this is already quite a lot i don´t need my swapfile to be as big, i need it just to be there in case some MB need to be shifted to the swapfile. I´m having the ram / swapfile ratio reduced to 1:1.

and even when i do some extreme calculations, i never even come close to filling the swapfile.
hope this is at least a bit helpful.

 

Wasnt he asking about the virtual memory shared with HD2600??

 

whoops, my bad.

should have slept a little more....

 

Im not sure, just a thought

 

Prior to the my using the Bios update I was having frequent blue screens and seemingly random window error messages. I won't bore with the details, but while I am waiting on a new machine I down loaded the new Bios package.

In limited run time since the update, no blue screens and only a couple of non-fatal error messages. The machine is still not perfect but much more stable. I was surprised since the mother board & graphics card had been replaced to no apparent effect.

 

Just goes to show you how useless tech support is half of the time They don't know any better than we do!

Hopefully you get a new machine soon Arthur and the update fixes everything--you've been through laptop hell and back it seems :\

 

It very true. Some techs will advise and order 2 parts like it made a difference ans so on, while some will only change one...

 

guys im having alot of trouble OCing the 8510p
i tried using ATi Tools but the speeds would not stick so I resorted to using AMD GPU Clock Tool. moreover, i cannot seem to surpass 600 core 650 mem. Sometimes it works, but most of the time it doesn't and I get a white screen of death. does anyone else have the same problem.
btw, im using cat 7.11. and vista

 

have u updated the bios??

 

yes. does anyone else get a white screen when they OC too much

 

I get a black screen and a halt if I try a value that it doesn't like. Wierd, though, as it didn't like 775 (instant black screen and death), but runs at 801.

I'm pretty sure I reached the max this particular gpu/mem combo can do, and I'm not anal enough to piss around for individual megahertz steps. I tried in chunks, and picked a 60% value between one that worked and the next one that didn't.

My clocks before I got bored:


Everybody keep in mind, however, that when you go from something like 150fps to 250fps on an overclock, that's essentially meaningless. This is because 150fps is 0.007ms/frame, and 250fps is 0.004ms/frame. That's a 0.003ms difference.
Something like a 30fps to 60fps is a bigger improvement, even though it's only 30fps rather than 100fps change, because the elapsed time between full frames is much greater.
Most of you probably already know this, but when overclocking comes into play, I see way too many people on the intarweb become overzealous about increasing their framerate by 100fps when they're already damn high. It just doesn't make a difference unless you're already damn low to begin with.

 

Great score sordid! Ive seen something similar with o/c on the forum where people shared their o/c experience with the normal cards, especially HD2600XT. It kind of didnt like o/c below certain value, yet if it was set like 20MHz higher everything was ok funny but it worked 4K barrier has been broken than
if only the crashes, and other problems with that gfx were sorted out I will be happy again

 

interesting to see how the ratio between core and mem clocks have such a different effect to the 3dmark06 results.
at some point during ocing the core clock the result won´t become that much better unless you make the ram a little faster too.

@ destiny

the 4k barrier has already been broken :
http://forum.notebookreview.com/showpost.php?p=2794686&postcount=24


//edit

@ sordid

well i´m curious, since your settings for core clock were 641mhz/801mhz and you have reached "only" 4090 points ( sm2.0 1434 / sm3.0 1794), which is a pretty good result by the way.

i have reached a score of 4143 points( sm2.0 1430/1759) with core and ram clocked to 636mhz / 687mhz.
of course your cpu is a bit slower and this maybe making your result a little worse, but when i compare the shadermark results for sm 2.0 and sm 3.0, which both are mainly affected by the clocks for core and ram, i see about the same result for both our clocking settings.

but since you have clocked the ram more than 110 mhz higher than me with having nearly the same setting for the core , i come to think that there must be a certain ratio for core and clock.

In the case of your clocking settings , your ram is fast, but it won´t get enough data from the core to make use of that extra speed. And i guess you could even clock the ram much lower, without loosing a serious amount of 3dmark points.

 

3D mark is not a very good comparision... it would be best to see a difference in gaming performance between these two settings. something like Crysis would be good. or Lost Coast.

 

At 641/693, I've attained 4010 (1407/1746/1782). I trust that the rest of the difference between my score and yours would be most likely CPU score, a slight increase in SM2/SM3 scores because of CPU as it's not 100% dependent on GPU, and standard error. Man, the core is one helluva bottleneck.

crysis gpu benchmark:
500/600 - 24.51fps avg
641/693 - 31.13fps avg

I noticed geometry dancing in and out of oblivion during the 641/693 test, but I also noticed some sections of ground doing wierd things at the beginning during 500/600. I cannot recall if there's a HD2600-Crysis issue or not.

 

3651 @ 600/700

Idle GPU temps are in the 30's, under load its mid 40's. If anyone remember i had concerns about my CPU temps under Everest Ultimate being too high. I think they are being reported wrong. Under speedfan under sensor core 1 and core 2 they are much more normal. Idling in the 40's then heating up accordingly. I have a bunch of temps that dont move in speedfan, just stay at the temp shown. Ackeron are you using amd gpu clock tool to measure GPU temps? If not try that cause i think they are accurate.

 

hmm, it´s really strange to see all these different temperatures ....

i measured my hd 2600 temps with rivatuner, amd gpu clock tool and ati tool.
all three of them do show the same values.

only speedfan reports different temps for the temp02 that is supposed to show the gpu temps.

it´s really weird, to see all these different temps for the very same piece of hardware. when i look at the temps bharris provided i still see a difference of around 10 °C even when i oc my card to it´s maximum of 640/750mhz

just a thought :
if i recall this correctly i think the hd 2600 is manufactured in 45nm process, is that correct ?
Is it possible that the change to this process reduced the core temps to these really low temperatures we are seeing here. and since this 45 nm process is new, are there good and bad cards around ? ( concerning the temperature)

 

What i find odd is my cpu temps, especially ackerons. The lowest im able to get my cpu is 43/44C after cleaning out vents, etc. I really do believe my gpu is being reported correctly as everest and amd gpu clock tool report the same and i see it rise when 3d apps are run. I did see my gpu get up to around 58 after playing around with o/c'ing running 3dmark06 over and over, but normally on idle, it sits in the 32-37C range.

Edit: Just when i thought my cpu temps were correct or so i hoped they were by speedfan, i dont believe they are anymore. Everest was reporting on idle 61c (with the cpu throttled down), which i thought was wrong. So i downloaded intel TAT and coretemp. They both show the same temps as everest. When im doing heavy cpu stuff like rendering, the cores can get over 100c, i think i may need to contact HP.

 

the HD2600 is manufacturered using the 65nm process, not 45.

even the desktop ati 3870 only uses a 55nm process.

 

ok, thanks for the info mate.
how about your temps ? these temp-readings for the hd2600 somehow don´t make sense.

summarizing the info on the core temps for the hd2600 in this thread

idle : 23°C - load : ~ 35°C - me

idle : mid 30´s °C - load : mid 40´s °C - bharris

idle : 23°C - load : ~ 31°C - vpicbm

idle : 35°C - load : 60-70°C - jerry66

idle : xxxx - load : ~ 47°C - shambles.ch

idle : 55°C - load : 70-75°C - ackeron

//edit

idle : 23°C - load : ~ 34°C - sordid //

as far as i´ve researched:

the one´s with the lower temperature were measured with everest, amd gpu clock tool or rivatuner. All of these programs do show the very same temperatures.

the higher one´s not only ackerons, but also the results from others using speedfan do show higher temperatures around 70°C.

I actually would like to believe the core temps for the hd2600 are really that low, but this just cant be true.
since the hd2600 mobility is quite new, is it possible that the asic for temp monitoring is different from previous generations? could that be a reason for these rather strange temp-readings ?

 

I'm pretty sure mine and schoko's temperatures aren't correct. The temps are ridiculously low. No computer hardware runs that cool, especially not an GPU.

I don't mind it though. I'll just make sure there's good airflow whenever I OC. Maybe I'm gonna buy one of these cooling pads, just to be safe.

 

I believe my temperatures are *reported* as about 23C idle, 32-34C load. I would sooner believe I actually experience jerry66's idling numbers though. A 65nm card will idle fairly cool, especially if it's mobile.

The heatpipe is probably installed by hand. maybe it was put on in a goofy manor at the factory.

-shrug-
I was engaged in a grueling 3 hour Battlefield 2 session yesterday, and not only did the body of 8510 stay relatively cool, but the air exhausted didn't heat up that much either. I know this because it was being exhausted right onto the mousing hand of a friend sitting to the left, and while he said it was warmer than ambient, it wasn't uncomfortable at all. my secret: I had the rear end propped up about an inch off the table.

 

I'll try installing Everest Ultimate and see what temps I get. But I dunno, just seems so weird

What other temp monitoring programs are there? Speedfan, Everest, AMD GPU (?), etc? I might just install 5 or 6 and see what results I get.

 

this is VERY good news my 8510p will be delivered tomoro morning !!!!
i dont like the temperature too high, so i guess simply up date the BIOS and no OCing will improve performance?
how much does just updating the BIOS improve?

 

@ bennyx8903

have you read this thread from the beginning ?
there were several issues this the new bios does fix.

1. corrected the clocks from 300/400 to stock clocks 500/600
2. ocing is possible now.
3. the vpu recover crashes and other crashes related to the graphics card are fixed
4. the fan does have a better management , does slow down after using 3d apps, which was an issue before.

@ ackeron

i´m curious what temps you will get with other programs.

 

Since i was having no problems with the previous bios, and dont have a need for o/c right now i reverted back to the old one. My temps across the board (both cores, GPU) have dropped about 10-15 degrees on idle. Fan runs quieter as result which is nice.

Very weird.