*Official ASUS W90Vp Owners' Lounge*

I'm so pissed off, I have w90 in my house and can't use it, because it's been USED already. Monday!!! They going to hear about it.

 

what do you mean used?

 

Wow thats a bummer hadis what happened ?.

 

I have never really understood why everybody seems to want higher 3D Mark scores other than for bragging rights (and I guess that can be pretty important) but the fact remains for me that flashing the GPUs and upgrading to CCC 9.4 gave me a noticeable increase in my actual GAMING performance which is all that I really care about. Jamiee, are you getting better performance out of your games? Is everything lower or is it just Vantage? Just curious I mean no offense by this post whatsoever.

 

kind of why your getting better performance now flashing and using the modified 9.4...because without people wanting to get more out of the system..you might be still stuck with the sucky scores and lower game performance when this thing first came out...(speculation of course) and vantage mark is more of true gpu benchmark than 3dmark..

and just like folks brag about ones football/baseball/soccer/hockey/basketball/swimming/track and field teams..... people will brag about ones 3damrk/vantagemark/gaming benchmark/game runs/fps per game/per online game and so on....

im just taking a stab in the dark on that though.

 

The main reason i want my machine to go/be faster is simply because im a gamer and in the future if i can squeeze an extra 10fps out of a game with tweaking,overclocking etc, then i just made an unplayable game playable and that is worth the effort and time spent . plus at times it can be fun though im not the bragging type. i just want solid and fast and dont care if someone has faster because im not rich and cant keep up with the joneses .

 

Mmmmm... That was a yummy dinner
Anyways,..
Sure bragging rights are always fun, but for me it's more just messing around than anything. I fully expect someone to blow away my Vantage score and that's cool.
As for gaming, yes the 9.4 Cats are an improvement. No question there.


I suspect the GTX 280 will be the new top dog in the right machine... but at what cost?
The W90 is a very nice machine at a decent price and in real game play probably won't lag much behind a good GTX 280 setup.
To each their own

 

yeah, that gtx280 is still up in the air at this point and the only gtx260 machine floating around is running xp..still waiting on that one to hit a vista machine....

but as of right now...there is nothing more fun than tinkering with ones machine... (big chili cheese grin)

 

Here an interesting comparo for you guys:

Laptop: W90Vp-X1
CPU = C2D T9600 @ 3.29GHz
Multiplier = 10.5
FSB = 1253MHz
Memory = DDR2 781Mhz
Divider = 4:5
Video = 4870x2 @ 600/900 ea.
Vantage GPU Score = 10943


Desktop PC
CPU = C2D E8500 @ 4.16GHz
Multiplier = 8
FSB = 2080MHz
Memory = DDR2 1040Mhz
Divider = 1:1
Video = GTX280 @ 615/1107 (Single card)
Vantage GPU Score = 11046

 

I have to test John its the only way, my gut says a few things.

1.) I think this untrue especially the 1280x1024 performance vantage run, res is too low for two high end cards not to be cpu limited.

2.) No because thats why my cpu score is killing the dual core guys at 2.75 gpu vs there 3.2+ ghz. My gpu score has been lower so far than everybodies with the dual core but my overall score higher due to the boost in cpu score.

3.) Probably due to the bottleneck, if you throttle it down below that point your going to see points start going away, the easy way to know this is when gpu-z supports "load" sensors if the cards are not at or near 100% load then they are not being used to there fullest.

4.) Depends on the situation and program/bench

5.) The system fsb is already way faster than anything the system can be doing its a huge bandwidth like Sata II is for hard drives. Its way faster than the hard drive is. Raising the FSB is only doing two things, overclocking the Ram and the CPU the FSB incrase itself is doing nothing. The ram overclock can very much help in some situations like when textures are swapping in and out of ram often. The cpu overclock is no different than if done by multiplier.

6.) This is what my tests are for, to show cpu/gpu relations and also I need to test the theory I saw mentioned today that my modded bios are lower than the stock ones.

 

Remember the ram on the W90 is actually underclocked at default CPU speed.
Even if I max out my CPU/FSB speed via Turbo Gear, the ram still never reaches it's full default speed of 800MHz.

 

Mine goes to 914mhz past the DDR2800 when I do the 2.75ghz. This is why my overclock is stuck at 2.75ghz ram wont go any further.

 

I've got a quick question. If you are limited by your RAM with the q9000 wont the same happen to qx9300 owners around the same mhz?

 

I got somebodys RMA'ed unit. Stickers glued back, can clearly see that they were pealed off before, Screen protector has lots of fingerprints, and stickers holding it also were preciously pealed, manual had pages folded, battery hole has scratches from previously putting battery inside. etc. It so obvious... I knew I would have some sort of bad luck with this, I told you Quad its gonna be me.

 

That really sucks man . Im curios was the tape on the box it came with blue ? with asus white letters.

 

It was, it is factory sealed with the blue tape and everything. They did very sloppy job repacking though. I just do not want a used unit. Going to fight with newegg tomorrow.

 

That's cos yer using SetFSB...

Is it just me or is this site slow and sometimes inaccessible tonight?

 

it happens all the time

we spam to much

 

Well Spam is America's favorite food... (Oh, wait..... You weren't talking about that kind of spam)

 

Ok I am back after over 2 long hours of watching Vantage run.

I have my tests, I have my numbers, I have my proof that what I assumed true before is true.

Vantage Performance 1280x1024 650/850

2.7ghz - 10255 gpu score
2.3ghz - 9544 gpu score
2.0ghz - 8864 gpu score

Notice the huge drop in gpu score by only reducing the cpu speed? This is a sure fire sign of a cpu bottlenck as to be expected of two high end cards linked together on a low resolution low setting benchmark.

Next we move it up to High resolution/settings this should show much less reduction due to cpu speed as more load is on the gpu thus less load on the cpu.

Vantage High 1680x1050 550/850

2.7ghz 6851 gpu score
2.3ghz 6709 gpu score
2.0ghz 6532 gpu score

There you go same cpu reduction as before but the gpu score did not move that much at all, still two 4870s in crossfire play most games at this resolution maxed out in the high 150+fps range easily leaving a cpu bottleneck. So there is still some but not much of a change.

Last I did my own custom test 1920x1080 all settings maxed to extreme

2.7ghz 550/850 - 3098 gpu score
2.3ghz 550/850 - 3003 gpu score

I stopped there I saw what I wanted to see, by maxing out the benchmark with the highest settings I saw no change to the gpu score as I have gotten to the point where the cpu was not a factor. the 95 points I consider within margin for error.

So now what I need to test is to see if we are now in the realm of extracting the gpu power via overclock!

Here you go

Vantage Custom 1920x1080 all settings maxed to extreme

2.7ghz 610/955 - 3393 gpu score!

Almost a 300 point gain! This shows that the overclock is working and it made a big difference now that I took the cpu out of the equation, it also shows that my modified bios is working.

I only have one more test to perform, I am going to flash my bios back to the original none modified ones to see if the score is higher.

I have a feeling it wont be and that the lower score reported was due to the overdrive not being on as every reboot it seems I have to go in there and mess with it to get the new clocks to stick.

Update

Vantage Custom 550/850 Original Bios 2.7ghz gpu score - 2893 (lower than the score for my bios but within margin for error)

So that settles it.

Now I will quote john and and state answers again not based on what I knew already but on what I tested. (lol John if that sounds harsh its not, I did know this already so thats why I had to take the time to prove it)

1: there is no cpu bottleneck
- Wrong there is, and its very very clear from the benchmarks the lower the resolution and settings the higher the cpu bottleneck, same way it always has been. A higher load on the gpu produces less frames for the cpu to have to process and thus less load on the cpu, if a benchmark is too easy for a gpu it can easily overload the cpu with frames and thus a bottleneck is formed.

2: you will get the same exact cpu score if you run 1 core vs 4 cores.
- This test did not show this to be wrong, however you can conclude this based on my previous full benchmarks that included the cpu runs, my quad @ 2.75ghz had a much higher score than the X1 users well over 3ghz showing the extra cores do matter, this is also why my total vantage score has been higher than any X1 user despite there gpu scores being higher.

3: your gpu score will take a very little hit when going from it's low level speed to it's high level speed
- CPU bottleneck is the problem when I maxed out the benchmark the small overclock gave us big results over 300 points more when our score was @ 3000 thats a 10% increase in performance going from 550/850 to 610/955.

4: when over clocking using just the multiplier it will not change much either.
- The results show the score continues to grow with higher cpu speed, it does not matter how you get the speed its going to help for benchmark reasons, real life though where 60fps is what your targeting it wont really matter.

5: when using setfsb to raise the system speed, it will start to change allot.
- This was your debate saying that it was not the cpu speed giving me better gpu numbers but the fsb, this is wrong. the system FSB is lightyears faster than it needs to be to handle all the information paassed through it, overclocking the fsb has no effect on performance. The ram being overclocked increases performance butt on a scale so small you would need to run very specific benchmarks to see it and will not see it in real life gaming.

The proof that its the cpu speed and not the fsb can be found by cross referencing my tests with others. The standard performance level vantage test
My Scoree
2.0ghz - 8864
2.3ghz - 9544
2.7ghz - 10225

Now Jamiee's score @ 3.3ghz 11009

Math time. My score jumped about 600 points per overclock and each overclock is about 300mhz.
Jamiee has a 3.3ghz overclock this is 2 more jumps over my 2.7ghz result @ 300mhz incriments. So with very rough math you should add 1200 points to my top score of 10225 giving you a number of 11225 almost exactly what jaamiee got! However his cpu multi is much higher than mine, so his FSB was LOWER than mine when he got that score.

Proof its pure cpu speed/cpu bottleneck.

6 you only need to attempt to over clock your gpu and leave everything else alone. since with a d901c. that is the only thing being over clock. so no it's not going to be the same thing. (jamie is over clocking his whole system to get where i got with only gpu over clocking) gpu over clocking with nvidia cards (9800m gtx)nets like a 3000 point increase (vantagemark) where as ati doesn't. this is why we sit and wait for the m980nu to come out fully. this is the only machine running dual 9800m gtx or higher nvidia cards that may have cpu/fsb over clocking then we can see what happens.
you have to look at the bigger picture. this is why we are trying to push the w90 to the upper limits now. because if that other machine is allowed to do all these things..it will leave the w90. 20k 3dmark score or not. now if it's released with no over clocking and a locked pll...then it's a wrap.

- CPU Bound

 

Just an FYI, the 60GB OCZ Vertex SSD is on sale for $190 at Newegg right now. I know you all are big time enthusiasts so I thought I'd post here.

See more details in the tech bargains subforum.

 

Very nice work@vicious

 

Those numbers are significantly lower than what I saw from other W90 owners, unless I'm missing something.

 

most of us are far more interested in a qx9300 right now but thanks for the heads up.

 

"I have a feeling it wont be and that the lower score reported was due to the overdrive not being on as every reboot it seems I have to go in there and mess with it to get the new clocks to stick."

I've noticed this as well. The 2nd cards clocks will usually stick, but the master's doesn't always. Sometimes even just after setting a new clock and saving, it'll revert back to default without me knowing.

Kind of annoying really..

 

these are gpu scores only not total numbers from running the whole test, also my numbers are low than the higher mhz dual core x1 owners for the reasons I stated, the cpu bottleneck.

I just rebooted and flashed the bios. I have one more run to do (thank god came to work 2.5 hours early just to do this for you guys and finalize the debate)

If this run has a higher score than my last custom run @ 2.7ghz 550/850 then we know my bios is lowering the score for some reason. Be back in 10.

 

terrytown again

http://forum.notebookreview.com/showthread.php?t=370989

 

I edited my last post a page back to end my test results and put to rest what john told me

Now I have one test left to do tonight, another card bios mod soft modding the voltage of the cards to see if we can get higher overclocks.

note that based on my long line of testing I just did that to get any benifit form a gpu overclock you going to be needing one hell of a demanding game all maxed out, or a very very fast cpu.

The first will help you in real life, the 2nd will only help you in benchmarks.

 

Seems ur right about the bottleneck because XoticPC got similar scores with the Q9000. But wow, I guess it's the same thing of ATi cards being influenced by the CPU speed regardless of FSB pumping.

We would need to see the GPU scores of a W90 system with a QX9300 to see if just the stock CPU speed would put the GPU score between a T9600 equipped system and a Q9000 equipped system.

 

Its just benchmark world, in games almost every game is playing the same. Im happy with the Q9000 though I do want a QX9300 down the road.

While the forums were down I created something for everybody

I created and tested my new gpu bios.

This one does two things increases the overdrive liimit to 800/1100 AND increases the voltage sent to the gpu.

This increase in voltage means higher gpu clocks are possible, but also increased heat, and higher chance of card damage so I recommend it for extremist & benchmarking only.

When I say temp increase I mean it, my max temp under 100% load before this mod was 74c, after this mod my card hit 80c and it was the first time ever I heard the W90 fans kick into high mode. Its good to know that the w90 is capable of great cooling and has even faster fan speeds if needed.

80c is a safe temp for the cards though. Id not want to go above 90c.

Still I recommend you maybe pick up one of these: http://www.youtube.com/watch?v=Ox2eaBjDfEg&feature=channel_page

A Vicious Super Mega Cooling Pad v2005

I got my card up to 665/1000 and it started to artifact on me a bit. I did not test my secondary card.

The new bios if you want to use it is here: http://rapidshare.com/files/220675429/W90_Limit_Breaker_Bios.zip

 

X-25M's work great out of the box, no OS tweaking needed.

 

I was actually thinking if that CPU bottleneck would probably explain the Crysis issue at DX10 you were having. Probably more tests may prove or refute that, but the Q9000 is good enough and can turn into a beast with setfsb.

As for the new GPU BIOSes with the upp'd voltages, 80 degrees C is fine, though I agree it's for the guys who want to go hardcore on their W90s. But with these new revelations, I wonder how long it's going to take for some crazy lunatic to get his or her hands on a W90 and a QX9300 and these BIOSes and just max out everything, until another lunatic with a M980NU does the same if its limits can be cracked like the W90s.

We're still trying to crack open the W840s here. We just need low latency DDR3 ram at 1333mhz or higher and maybe some new gpus too.

 

Many thanks sir, i know ill be needing the new bios once my laptop gets in. Going to go drool over it today and get a real feel for its gargantuan size. rep coming your way.

 

Yep I already stated Crysis was the only game so far that the q9000 is doing worse than the dual core and showed signs of the cpu bottleneck.

Its just easier to believe now with testing to backup and not just my word. Still 2.75ghz should be more than enough, thats bad coding at work for you. I wonder if its even using 2 cores... It may only be 1.

 

If you are hitting a wall when overclocking due to the ram wont a QX9300 do the same? Sorry if this is a noob question, but if the qx9300 wont hit this same wall as the q9000 can you explain why please? Thanks

 

yes I will hit the same FSB wall but the qx9300 has a cpu multi 2 higher than the q9000, my limit right now is the ram, not the cpu. With the new cpu in place the ram will still run the same speed. it has its own "divider" the W90 uses a 4:5 FSB:RAM divider.

So my current overclock of 366mhz puts the ram FSB at 457mhz aka DDR2 914 as DDR is "double data rate" and runs twice as fast as the FSB its using.

The Q9000 Multi is 7.5 the QX9300 Multi is 9.5

So my current overclocck of 2.75ghz uses a 366mhz fsb.

366 x 7.5 = 2750 aka 2.75ghz

that same fsb overclock with a qx9300 in place

366 x 9.5 = 3477 aka 3.47ghz

If all goes well and I get a good cpu (as all extreme cpu seem to be) I can expect about 3.5ghz out of that cpu, basically a multi 2 higher means adding what ever fsb I can get two times + what I have now.

3.4ghz is a big jump in terms of video rendering, and in our current state for benching thats enough to remove the cpu limitation for some of these benchmarks.

The higher cpu multi only gets better as you overclock higher, if I got new ram that could do 400mhz fsb

400 x 7.5 = 3ghz
400 x 9.5 = 3.8ghz

Plus every bit of overclocking on a quad is doubling up what a dual core can do in terms of raw power.

for any program that can use 4 cores (pretty much any program that uses the cpu as its primary rendering tool) a 3ghz quad would run so fast that you need a 6ghz dual core to run the same speed.

This is why I much prefer the quad core even at a lower speed.

 

Ah, thanks for the explanation =]

 

vicious you are doing alot of typeing and good work on that new bios ... i need one for my GPU next

 

i was going to draw up a long reply to all that, but im gonna let that go. ill some it up in a few words
1: cpu mhz drives ati cards.(been saying this since the 3870s) (if you call that a bottle neck, that's fine)
2:w90 is an under clocked machine untill you use turbo gear to put it where it's suppose to be.
3:when you raise the MHZ(fsb) which raises your cpu&ram and whatever else it's connected to, your ati cards run faster. sorry, but you never did a multiplier test. (only the cpu speed changes and not the whole system)(i did these test with a 2.4 ghz quad and multiplier only)
4:now take what you just learned and apply that to lets say..my machine. how far ahead of you do you think that would put me? since i have none of those things?and im right about even with jamie(for the time being)

and the whole part of this you missed...im doing a gpu over clock not a cpu over clock. so i guess going by your numbers and system..this system would be capable of doing 20k, since im bottle necked at the cpu. going by your math assumption there.

we already had the debate about which one was better long before and that was a cpu over clock using some form of fsb over clocking. you used it on your desktop. i don't think your multiplier over clock was going to help you at all.

 

win 7 is actually SSD optimized. i went through the check list to optimize for vista and looked through win 7 and most of it was already set

 

1.) cpu mhz drive any card, from ati to nvidia to no name since the dawn of video cards eons ago, how much is needed to drive them is dependent on the card and what your rendering, the point is that a high resolution & high settings benchmark/game is not going to require as much, you can get to the point where I was by maxing out vantage that the gpu becomes the bottleneck. overclocking the cpu higher would have not raised my score, thus why when I lowered the cpu speed the score stayed the same.

This is when overclocking the gpu raised the score. You have to learn where your cpu/gpu balance to find that bottleneck point. It of course will vary with each and every game or benchmark.

2.) no the fsb is where it is supposed to be or the cpu would be underclocked. The RAM:FSB ratio however is reduced by one to allow a safety net for overclocking with turbo gear. This only means the ram is not full speed by default and this has VERY little impact on performance.

3.) more cpu power allows the card to run faster if they are capable of it, vantage performance run for instance the cards can run very fast as its too "easy" for them, but maxed out vantage the cards are not able to speed through it, it doesnt matter how much cpu speed you feed it past the bottlneck point (lower than 2.3ghz) they wont get faster.

I dont need a multi only test to draw conclusions, I know how this stuff works been doing it for years and years. I have done multi only changes on my desktoip and ram multi only changes and saw how the performance changes based on both indpendently so i know whats going to happen when i change them together.

I purposly dropped the cpu multi on my q6600 down by one for a higher fsb & ram speed, an extra 100mhz on the ram and a 1:1 fsb:ram ratio didnt even give me 30 points in 3dmark06 or .002 seconds in wprime, the change is very small.

So for all intentive purposes a fsb overclock = a multi overclock

Back in the P4 days ram latency timings and fsb overclocking did make a more noticeable difference, these days its more of a bragging right.

 

just do the simple thing and run just an over clocked gpu. that's it thats all... thanks

edit: and for the record..everything has a bottle neck...if it didn't...then we would be at infinity

edit: ichime
what happens if you try to lower the timings on the 1333 memory. does this have any effect on stabilizing it?

 

I already have, what are you looking for that I didnt already do?

 

Its also a good brand of speaker and a car

 

pshhhh they suck lol

 

Na I loved my Infinity Perfects, some of the best thief/small pararmiters to be had on a speaker in that price range.

Great combo of SPl & SQL, won a few armature comps with them. Also I prefer speakers like that wich function in a small sealed box, takes up less space and better tonal curve than ported boxes that have to be tuned for a specific frequency peak due to rollover.

If pure spl is your thing yeah the perfects are the wrong place to look, but I want to listen to my music not just make noise

 

I have my tests, I have my numbers, I have my proof that what I assumed true before is true.

Vantage Performance 1280x1024 650/850

2.7ghz - 10255 gpu score
2.3ghz - 9544 gpu score
2.0ghz - 8864 gpu score <--- this one. now over clock the gpu higher than 600/900 and report the results (example 650/1000 or higher)
then post what you get. then we will take the number of your low end gpu and compare it to your over clocked gpu

Notice the huge drop in gpu score by only reducing the cpu speed? This is a sure fire sign of a cpu bottlenck as to be expected of two high end cards linked together on a low resolution low setting benchmark.

Next we move it up to High resolution/settings this should show much less reduction due to cpu speed as more load is on the gpu thus less load on the cpu.

Vantage High 1680x1050 550/850

2.7ghz 6851 gpu score
2.3ghz 6709 gpu score
2.0ghz 6532 gpu score

There you go same cpu reduction as before but the gpu score did not move that much at all, still two 4870s in crossfire play most games at this resolution maxed out in the high 150+fps range easily leaving a cpu bottleneck. So there is still some but not much of a change.

Last I did my own custom test 1920x1080 all settings maxed to extreme

2.7ghz 550/850 - 3098 gpu score
2.3ghz 550/850 - 3003 gpu score

see where i have it in bold up above.

infinity is a nice sounding system though

and infinity is infinite. meaning on going never stopping. ( exaggerated example: we have no bottle necking then what is stopping the system from going higher and higher)

definition:
Infinity
Unlimited extent of time, space, or quantity; eternity; boundlessness; immensity.
Unlimited capacity, energy, excellence, or knowledge; as, the infinity of God and his perfections.
Endless or indefinite number; great multitude; as an infinity of beauties.
A quantity greater than any assignable quantity of the same kind.
That part of a line, or of a plane, or of space, which is infinitely distant. In modern geometry, parallel lines or planes are sometimes treated as lines or planes meeting at infinity.

why i say exaggerated



but still, all great stuff you posted! very informative to say the least.
but nvidia aint cpu driven, they can drive them selves.

and Bose is probably better than infinity now...speculation of course.

 

Now what is infinity times i? lol

Anyways you will see a very small boost in score if I overclock the 2ghz cpu bench.

I'll explain why.

when you run a benchmark like vantage it is not a constant load, you will see say 20fps in one scene and 60 in the next. If you had a monitoring program showing gpu load it would be fluctuating.

When the fps is very high this is when the cpu is most probably going to be bearing the most load and having more chance for limiting the system. however there will be points where the gpu may reach 100% load for short periods of time due to something incredibly hard to render on screen.

Those short & small moments are now going to be able to run faster if you overclock the card as what was 100% load then (before the gpu overclock) is now only maybe 95% load and the new 100% load will be faster.

However in that low res performance test those moments are going to be incredibly brief (actually in performance there may be none) but this is why running Vantage Extreme, the same gpu overclock is going to show a much larger boost, as there will be many more times the gpu was at its limit and it needed room to grow.

performance is not demanding enough on the gpu without excessive cpu speeds for an overclock to do anything.

In a nut shell it boils down to this:

if the scene being rendered is too easy for the gpu the cpu is going to be the bottleneck being at 100% and the gpu will not be able to reach 100% as the cpu will hold it back. In very difficult scenes where the gpu struggle it will be 100% and the gpu is the bottleneck and the cpu wont be maxed out or have much to do, its basicly going to be waiting for the gpu.

something low like performance is going to be almost entierly cpu bound, while something high like extreme is going to be mostly gpu bound.

in the performance run the cpu will be maxed out load most of the time, in extreme the gpu will be maxed out load most of the time.

When ether of those is at there limit to increase your performance higher you need to overclock that limiting agent, so for performance runs you need to overclock the cpu to see the big gains, for extreme you need to overclock the gpu for gains.

however due to the fluctuations you will have small moments where the bottleneck point changes so you can get small benifit from increasing the limit of ether of the agents.

 

It seems to me like the both of you are basically saying the same thing; at low res, the cpu is the bottle neck. At high res, the gpu is the bottle neck.

CPU overclocking will increase graphics performance more at lower res and has less influence at higher res.

GPU overclocking will increase graphics performance more at higher res and have less influence at lower res.

Regardless, since complex games are not 100% about graphics power, a higher cpu should always make some difference in game, regardless of the resolution.

...at least that's what I've taken out of the discussion so far.

 

In a very simple way thats part of the foundation of how it works.

higher res is harder on the gpu so thats a prime factor.

games that stand out for needing cpu power FSX & Supreme Commander, due to the physics & AI eating up the cpu power. (both quad core games too)