Underclocked MBP

I heard that the graphics card in the new MBP is underclocked. What exactly does that mean? I kinda know, I think it makes it slower or something. Also, how do you fix it to not make it underclocked?

 

Underclocked means that the core and the memory of the graphics card is being ran at a slower speed then the reference speeds. l believe apple does this because of possible heat issues. You can always change the clocks to whatever you want using a 3rd party program called ATI Tool.

 

A) You can not change the clock speeds in OSX ... no tool exists (yet) to do so.

B) ATI Tool doesn't work in Windows using Boot Camp. I think PowerStrip does, but the results aren't very stable.



-Mickey

 

Hi I don't know about macs but check my sig. You may get an idea.

Cheers,

 

I am using ATI Tool in Windows on my MBP, it works great. I'm running at 406, 400, no issues. I turn it back to default when I'm not gaming though. I get around about 3800 in 3dmark05 with these settings, and gaming is great.

Stock settings were 316mhz core, 297mhz memory. This card runs is clocked at 470core,470mem in the similarly spec'd Acer travelmate 8204wlmi, but the macbook has poorer ventilation and a thinner chassis, that is why it has been 'underclocked'. By 'underclocking' the graphics card, the card uses less power and generates less heat. Hopefully that answers your question.

 

AIBs or OEMs decides the final clockspeeds on every notebook after their thermal design. Even though several notebooks all have, let's say, an Mobility X1600 does not mean that they all fit within the same thermal design.

Hence, the clock difference either from model to model or from brand to brand.

In other words, the Mobility X1600 is not underclocked but rather clocked to fit within the thermal design of the MacBook Pro.

 

This is what bugs the **** out of me about the underclocking. Regardless of what you "think", ati has specs and the macbookpro's x1600 falls FAR UNDER those basic specs. In fact, the macbookpro's x1600 barely beats normal clocked x1400 systems, and can lose to x1400 overclocked laptops. Why did they use a more expensive video card, then gimp it to death? The obvious answer is the keep heat undercontrol, but why didnt they just stick a x1400 in it? Almost the same speed, and would be cooler and cheaper.

 

To be able to advertise that the MBP comes with an x1600. Not many people know that it comes underclocked. They see x1600 and they say, "oooooh, ahhhhhh."

 

The underclocked x1600 is significantly faster than the standard clocked x1400,
somewhere in the ballpark of 50% faster by most benchmarks I've seen.

Comparing to overclocked x1400s is irrelevent for product design, so apple would
not consider overclocked x1400 speeds in such decisions.

 

You seem to fail to understand the concept. When AiBs or OEMs buy chips from ATI, they may clock them at whatever they see fit. Again, the Mobility X1600 found in the MacBook Pro is not underclocked in any way. It is just clocked differently. ATI is not the one who decides the clockspeed of these chips. They give safety margins in which the chip works.

 

To further Pressure's point, if an OEM buys a large lot of some part they can get an
additional price break if it has a lower operating specification window. So ATI may
sell lower specificiation x1600s to Apple at a lower price...maybe even a lower price than
high specification x1400s.

In CPU-land, many chips are various speeds come off the same line and are only
differentiated by post fabrication tests. I am sure this applies in GPU-land too, so ATI
almost assuredly have inventory of the same chip with various safe speed windows.

This easily explains why only some people who 'overclock' their Apple x1600s can
reach the x1600 clocks in 8204s or W3Js...and almost nobody can get an Apple x1600
to clock at what the x1600 in an 8204 or W3J can overclock to.

 

The only problem being that there is no yield issues of the RV530 core at TSMC.

The chip is fairly small too, only 157 million transistors. For comparison the X1800 has 321 million transistors.

I think the main difference is the voltage supplied to the graphic chip and that should explain the lower clock (also when overclocking). Not to mention make it fit the Thermal Design of the MacBook Pro chassis.

Although it would make sense if they speed/voltage binned them and had a fair deal that didn't reach the desired speed. Better making money off them, than throwing them away. But they would need a steady supply of "rejects" for this to make sense . . .

 

Depending on their process, they probably do have a steady supply. If such supply ran
out, they simply use chips from the next bin up. This would just mean their process is
working better than expected and they have higher quality coming out. It does not cost
ATI more (unless their contracted fab charges ATI per quality delivered) if their
percentages of 'reject' chips is lower than expected. Actually, this up-binning is likely
done at the fab itself, and ATI probably does not even know or care when it is getting
better chips per level than contracted.

I'm not sure I'd call 157 million transistors small. Isn't the core duo itself somewhere around
152 million?

 

What ever makes you sleep at night. The fact is, ati sets acceptable specs and other companies use those specs but Apple does not. They gimp their x1600 to not even be on the same level as other notebooks who use x1600's. It is called underclocking, and many people do it with their cpu's and gpu's to keep heat down. Apple has done this. Yes, they are setting their gpu clock speed but by setting it so low, it is underclocking.

 

If you guys are curious about it, google up the heat issue and heatsink compound problems on MacBook Pro's. They overload them (insane amounts) with heatsink compound, and it ends up running the cpu 20-30C higher at idle, and 40C or so higher under load. People have been seeing 90-95C under load, and high 80's idle.

This is the cpu, the gpu's tend to throw off more heat, so you can understand why they might be underclocked a bit.

Here is a thread on the subject on another forum I am on:

http://forums.somethingawful.com/showthread.php?s=&threadid=1864582

 

Lots of other manufacturers do this 'underclocking'. Take a look at benchmarks for other mobility cards. Ever wonder why different notebooks with X600 can get anywhere from 900-1700 in 3DMark05 at their 'stock' speeds? The Asus Z71V has a badly underclocked 6600go but I don't recall too much noise over that and some people are still recommending them!

Back in the AGP days, ATI released the MR9600 Pro Turbo that actually had minimum clock requirements because manufactures varied the MR9600 clockspeeds so much. I don't believe ATI had any such requirements for any other GPU. Which explains why some notebooks with the MR9700, immediate successor to the MR9600, are slower than the Pro Turbos.

It sucks that the X1600 is low clocked in the MBP, but ppl have to stop with the BS that the X1400 is faster The MBP's X1600 still outperforms the X700 and kills the low end cards.

 

Let's say the clock was at ATI's default specs, would it be possible that the chip is underclocked (10-30% by user or manufacturer) on purpose to promote better battery life? Or is there also a certain "minimum" clock speed the chips have to run at before everything becomes screwy?

 

The underclocking of the 'Max Performance' clockspeeds won't affect battery life unless you plan to game on battery. I'm not sure if there's much variation of the clockspeeds when PowerPlay is set to 'Optimal Battery Life' among X1600 GPUs in various notebooks. I would think that ATI themselves would have set the clockspeeds and voltages as low as they think is possible for all the chips.

The problem with user manual underclocking is that currently, PowerPlay has to be disabled in order to change clockspeeds manually. From just personal testing with the X700 in my Z70Va, setting PowerPlay to 'Optimal Battery Life' used less power than disabling PowerPlay and setting clockspeeds in ATI Tool as low as they could go. It may be different now, since when I was testing I couldn't change voltages.

 

Hi lazybum,

I have a pretty much different experience. You may check here:
http://www.thegamebooks.com/powerplay-vs-underclocking-t146.html

There is also a graph showing the underclocking performance.

Cheers,

Ivan