display anomalies

Don't worry so much. I suggest that you try to reproduce the issue yourself. The issue is not due to a memory stress test but rather due to a graphics stress. Now it's something that is usually only produced when the device is under extreme heat stress and I doubt they would do that in the shop. So try to run a ton of programs at once and if you have the problem press Cmd Shift3 to get a screen capture. Once you have half a dozen of these go back to the store to show them.

Message was edited by: Mad-elph

I've been having this problem to some degree since day 1. (I have a Core Duo 2.0 GHz MBP with X1600.) I originally only saw issues occasionally when playing WoW -- I'd get graphics glitches after playing for a long time. Pointing a fan at my laptop seemed to solve the problem. Months later, I started seeing the horizontal line of corrupted pixels too. Later still, WoW became unplayable -- graphics glitches to the point of not being able to see anything in the game, and it was happening all the time. I downloaded the SMCFanControl and discovered that if I cranked the fans up all the way (6k RPM), the problem went away. Around that time, I started having crashes (even when just running Aquamacs or something and load on everything was low.) I eventually figured out that the crashes correlated with the horizontal line corruption. At this point, even with the fans cranked up all the way, WoW was unplayable -- graphics glitches and crashes following shortly after. Some of the time, there is an error message in one of the log files that suggested that the graphics hardware was at fault. The error message was:

ATIRadeon::wait for_rbspace: Overflowed block waiting for FIFO space. Have 4294967294, need 5.
(there is a hex dump of something after that)

I figured that the heat sinks had clogged up with dust, so I took everything apart and cleaned out the heat sinks, (they weren't very dirty after all,) and applied Arctic Silver 5 to the 3 chips after removing the old thermal material. I noticed that the submount PCBs under the 3 chips all appeared to be damaged. The upper surface of the submount wasn't flat any more, and there was discoloration. After putting it all back together, the CPU and GPU temperatures were slightly lower for a given load and fan speed, but aside from that slight improvement, I still see the same issues when the GPU gets into the 45-50C range. If I want to play WoW on this laptop, I have to have a big fan pointed at the laptop with the internal fans cranked up all the way, limit the framerate heavily, (to 25 FPS or lower), and I still sometimes get graphics glitches. This machine now crashes sometimes just running a browser and a text editor. Any time the GPU temp gets above about 45C, I start having occasional issues. If the GPU temp ever gets above 50C, a crash is imminent. There is something wrong with the X1600 chips or the video RAM in these laptops -- data errors should not be occurring in silicon at those temperatures. Also, the thermal interfaces between those chips and the heat spreader/heat pipe assembly can't be that good -- it is hard to make something that large with that many different mechanical interfaces flat enough to have a thin thermal interface. That must be why the Apple Service Manual for the MBP requires such a huge amount of thermal interface material to be used. Filling in the big (relatively) gaps that are probably present in one or more of those interfaces takes a lot of goo. Maybe the X1600s were originally fine and they've been cooked into disfunction by bad thermal design; it is hard for me to say why the X1600s are producing errors at such low temperatures, but Apple needs to fix this problem. It is one thing for a manufacturer to not repair something years after it was sold and worn out, but that is clearly not the case here; these machines had flaws from the beginning that just keep getting worse over time. What is the point of getting a machine with a powerful processor and GPU if the capabilities can't be used?

I think what really shows that this is due to a bad thermal interface design is not so much the underclocking, but the fact that the Apple service manual suggests that one should apply, I believe, 0.2-0.3cc of thermal paste to each pad on the heat sink, (one per chip). This is way more thermal material than would be necessary under normal conditions. I suspect the reason that they require that much is because they know that the thermal interfaces are going to be unusually thick because they basically have one heat spreader covering all 3 chips and it would be difficult to make the 3 chips coplanar enough that the thermal interfaces can be thin. It is like trying to set a plate on top of 3 soda cans so that the entire rim of all 3 soda cans is in good contact with the bottom of the plate. It is easy with 1 can, but when you go to 2 cans, if the cans aren't exactly the same height, your plate will end up just contacting a single point on both cans, and there will be a space between the rim and the plate everywhere else all the way around the rim. The same would be true for 3 cans. My guess is that the huge amount of thermal material is there to fill in the gaps. (That is what thermal grease is supposed to do, and why I put a large amount of Arctic Silver on mine when I put it back together.) In any case, such a thick thermal interface will have high thermal impedance, (thermal grease actually doesn't conduct heat very well; the purpose of it is to fill in gaps that are supposed to be microscopic between the die and the heat sink; the increase in effective surface area of the interface makes up for the poor thermal conductivity of the material, but the thermal impedance will still be much higher when the interface is thick like it must be in this design.) If I recall correctly, the interface thickness that is anticipated when putting a heat sink onto a normal CPU is something like <100 microns. (Just checked the Arctic Silver website -- about 75 to 130 microns.) This is why one normally can use such a tiny amount of grease. On a CPU die that is 10mm x 10mm, (100 mm2 surface area), a 100um thick interface would require only 10 mm3 of thermal material, (plus a little because it spreads out in a circle and some goes over the edges of the die). The 0.2-0.3 cc suggested by the Apple Service Manual is 20-30 times the amount normally required. They know those interfaces are bad.

Same problem here! (15" MBP 2,0 first gen, ATY X1600, bought in summer 2006) ... using SMCfancontrol eases the symptoms, but does not cure the problem ... whenever doing Grpahic intensive work like photo editing or CAD modelling ... the BP dies on me with exact the same glitches and problems as everyone mentions here!

<Edited by Host>

Add me to this unfortunate list. I'm without a computer now and have to resort to my iPod touch to post this. My video problem started last week while I was working in Photoshop. Small crosshatched squares started appearing everywhere and I immediately shut down my computer. I waited for it to cool down before I booted it up but that didn't help. The display at the boot screen (gray background, apple logo) was already garbled and the computer froze at the login window. I ran the Apple Hardware test for a result: failed it. I took out the battery and held down the power button to reset the SMC. I restarted after that and my laptop worked, great!

Well...

It's not working again four days after it started acting up. This issue is obviously not going to go away on its own. In fact, the problem is worse this time. My computer screen is garbled at boot up and freezes before the login windows during the blue screen (also a bit garbled). I'm going to the Apple store tomorrow to see what kind of comment the genius can muster up about this. I'm obviously pessimistic about this because of how widespread an issue this has become to such a specific video chip. I too also demand a fair price for a logic board replacement because my brothers HP bargain laptop is still living five years with a cat sleeping on the keboard every night.

I'll report back to this page with my store visit experience and maybe some pictures of my dead laptop.

Message was edited by: Hubbu

+"Also, the thermal interfaces between those chips and the heat spreader/heat pipe assembly can't be that good -- it is hard to make something that large with that many different mechanical interfaces flat enough to have a thin thermal interface. That must be why the Apple Service Manual for the MBP requires such a huge amount of thermal interface material to be used. Filling in the big (relatively) gaps that are probably present in one or more of those interfaces takes a lot of goo."+

I'm not sure exactly what you mean here with the difficulty of making an efficient heat sink system. I also want to clear up some misconceptions of the heat sink/thermal grease issue that others may have. The TG is supposed to be applied conservatively due to the fact that the thermal resistance is directly proportional to the thickness of applied grease. It's main purpose is not to solely transfer heat from the die to copper; it's supposed to turn an imperfectly flat surface into a "perfect" one, i.e. fill the gaps. Copper does a mighty fine job of transferring heat internally, but it has a tendency to not exhibit the surface smoothness required to do so at the efficiency needed for a computer. Ergo, the assembly manual (assuming that what we've seen is the official manual) is wrong. AMD even states that the amount required shouldn't exceed a grain or two of rice. Ironically, Macs are assembled in China. (no offense)

I think that's what you said, so I'm probably just regurgitating. Also I found some data/evidence to back up these claims:

page 8 has good pictures showing the purpose of filling the gaps:
http://www.amd.com/us-en/assets/contenttype/white_papers_and_techdocs/26951.pdf

the relation between thermal resistance and thickness:
http://www.microsi.com/packaging/thermal_grease.htm

I sent my MacBook pro off to an apple repair center. The Genius powered up my laptop and immediately said "oh wow." It's $310 to replace the logic board. I'm crossing my fingers that it arrives on time.

Hey all

I managed to convince Apple to replace my logic board for free, despite the computer being out of warranty. I called the tech people and they said there's nothing they could do; I politely asked to talk to the boss of the tech people, and he said there's nothing they could do; I politely but insistently asked to be transferred to the customer service department, and the rep I got listened to my story. I told her about the visual anomalies, the crashes, etc. - as well as the fact that I'd had to replace my hard drive and DVD drive, even though the computer was only two and a half years old. (The tech people actually told me that if I were a company, I'd only expect the computer to last that long!)

The customer service rep had me fax her the receipts for the replacement parts, as well as images of the fault, and then had me take my computer to an Apple service center, who confirmed the logic board problem, and Apple agreed to pay for the repair (as well as the service cost).

So that's at least one happy ending!

Add me to the list: iMac "Core 2 Duo" 2.16 20-Inch

Progressively bad horizontal lines, weird visual anomalies like color fields, blurred text, and geometric shapes of weirdness, constant freezing/crashing. Sometimes the screen will go black, but iMac still seems to be functioning.

Just tried the fan fix, so we'll see if that helps.

Apple needs to address this. For every person that has found this thread there are probably 5-10 that haven't. This is a big problem, and it needs to be fixed.

Hopefully, Apple will step up to the plate and address it.

Just got mine back yesterday--new logic board, heat sink, and fans. Last night it quickly went up to 70 degrees celsius and I started seeing the flashing pixels all over the place. I'm so frustrated, as you guys are also. I'm so disappointed that apple isn't doing anything to fix our machines permanently. I've always used a cooling pad--and am now increasing fan speeds w/ SMC. But it gets annoying and it is frankly ridiculous I need to do so.
All the best, K

Well it is now official.. My Macbook Pro 17, 2.33Ghz is DEAD. 2.3 years old.

Apparently,(this has been confirmed by Humac)it is the Logicboard = (*ATI Mobility Radeon X1600 graphic chip* )which needs to be replaced. Graphics are messed up. Started around april 10th 2009, with the screen just looking like a bunch of big pixel jumping around. *I WAS SHOCKED*. No warranty, so no free replacement for me. And there is no way im going to pay 1700 Usd to get it fixed. So my guess is that I will have to look elsewhere for a laptop replacement, unless apple realize what heatproblems these macbooks suffers from. Btw. where is the pro in that? I wanted the best computer to get the job done, but bought myself a tin can.

Good to see that the new unibody logicboard atleast look like it has been thought over. Thank you apple.

For the rest of you out there, good luck with yours.

R.I.P
Macbook Pro

<3 fred

Exactly the same thing here.

Because the laptop belongs to my work, they agreed to pay for its repair and sent it off. Apple have had it a month now and gave me a call last week to say they couldn't find a problem. Well, how about the fact that the whole thing freezes when it gets slightly warm? Useless.

onecoolirishlad wrote:
+"Also, the thermal interfaces between those chips and the heat spreader/heat pipe assembly can't be that good -- it is hard to make something that large with that many different mechanical interfaces flat enough to have a thin thermal interface. That must be why the Apple Service Manual for the MBP requires such a huge amount of thermal interface material to be used. Filling in the big (relatively) gaps that are probably present in one or more of those interfaces takes a lot of goo."+

I'm not sure exactly what you mean here with the difficulty of making an efficient heat sink system.

Normally heat sink systems are not hard to design or manufacture, but this one is different from most heat sink systems found in computers because the one in the MacBook Pro is a single, apparently rigid, piece that is meant to contact all three dies simultaneously. Normal CPU heat sink systems, for instance, have a single flat surface that contacts a single die and something surrounding the interface that puts a known and approximately constant amount of force on the interface. The single thermal interface in such a system means that only one mechanical datum is functionally significant in determining the thicknes of the interface -- the die itself. In the MacBook Pro system, there are three mechanical data to which the heat sink must be aligned, and these are all in nominally fixed positions relative to each other. (Of course the PCB can flex a bit, and the heat sink in this case may be able to flex a significant amount as well.) Considering that the three BGA packages are soldered to the logic board, and that their positions relative to each other are essentially finalized when the solder cools, if someone were to try to guarantee a thermal interface thickness similar to a single interface solution, they would need to control the tolerances of the positions of the heat sink surfaces relative to each other and the tolerances of the IC dies relative to each other such that the manufacturing of the heat sink assembly and the population of the logic board with the BGA packages would probably require non-standard techniques. I suspect that they thought they had shown a "good enough" interface thickness, and therefore, "good enough" cooling, using the method that ended up being used in our laptops. Perhaps they didn't expect that high die temperatures would cause errors as quickly as they have.

I also want to clear up some misconceptions of the heat sink/thermal grease issue that others may have. The TG is supposed to be applied conservatively due to the fact that the thermal resistance is directly proportional to the thickness of applied grease. It's main purpose is not to solely transfer heat from the die to copper; it's supposed to turn an imperfectly flat surface into a "perfect" one, i.e. fill the gaps. Copper does a mighty fine job of transferring heat internally, but it has a tendency to not exhibit the surface smoothness required to do so at the efficiency needed for a computer. Ergo, the assembly manual (assuming that what we've seen is the official manual) is wrong. AMD even states that the amount required shouldn't exceed a grain or two of rice. Ironically, Macs are assembled in China. (no offense)

I think that's what you said, so I'm probably just regurgitating. Also I found some data/evidence to back up these claims:

page 8 has good pictures showing the purpose of filling the gaps:
http://www.amd.com/us-en/assets/contenttype/white_papers_and_techdocs/26951.pdf

the relation between thermal resistance and thickness:
http://www.microsi.com/packaging/thermal_grease.htm

Right. Note that the examples in those documents assume a single interface heat sink situation, where the surface planes of the heat sink and the die will automatically settle nicely against each other when the proper amount of pressure is applied. That pressure-induced settling squeezes out excess thermal material, (although it takes time for that to happen,) and normally causes a thin, even thermal interface. As an additional bit of evidence in support of the "multiple interface heat sinks are a bad idea" theory, look at photos of the interface material left on the heat sink surfaces and the dies after the interface has been separated -- the interface is clearly thinner on one corner of the die than on the other corners most of the time. That suggests to me that the interface isn't very thin at all on average. Knowing that these surfaces weren't typically going to align as well as they would if they used 3 separate heat sinks, perhaps Apple tried to at least make sure there weren't any gaps in the interface by suggesting a big pile of grease be used on each one. If that is true, one wouldn't want to use a normal amount of thermal grease, because even if it isn't necessary to use a larger amount, a larger amount would insure that in the worst alignment case, there aren't any corners of the die that don't even have a wad of thermal grease serving as a heat spreader.

As another anecdotal data point, I have tried pointing a standard desk fan at my laptop while playing World of Warcraft, and while this initially reduces the reported temperatures of the CPU and GPU a small amount, the fans usually slow down to compensate for it. Strangely, it solves the graphics error problem though. Since the fan and heat exchanger parts of the heat sink are near the hinges of the laptop (on both sides) and the obvious place for the machine to draw air in is in between the exhaust areas, I wonder if the cooling system is drawing in heated exhaust air? Note also that the video RAM is kind of near the middle of the logic board, (those four largish chips, two near the rear edge on the right side, and two next to the right fan, all four on the top of the logic board when it is installed.) I noticed when I took mine apart that there is a thin black plastic film applied to the bottom of the top case. In the case of my laptop, this film was bubbling up off of the aluminum in a line all along the rear edge of the top case, (not melted, just the adhesive separating from the aluminum.) Considering how little space there is between the bottom of the top case and the logic board, I wonder if that film is effectively sealing off the space where air being sucked in through the middle of rear vent would pass over the video RAM, thereby causing it to overheat. I noticed that turning off my desk fan didn't cause the GPU or CPU temperature to rise because the laptop fans would speed up to compensate, but the graphics glitches would return with the external fan off even though the GPU and CPU temperatures were the same.
No matter how much I analyze this problem or try to fix or work around it, it still ***** that this expensive laptop is failing to perform normal functions when it isn't yet 3 years old. If the thermal design or the video RAM heat tolerance or the plastic film is the reason for the failure, a new logic board isn't going to fix it. I'm really beginning to suspect that Apple just didn't do accelerated wear tests or anything like that on this design.

jaylabelle wrote:
I'm really beginning to suspect that Apple just didn't do accelerated wear tests or anything like that on this design.

No ****!! MBP's dying after 2.3 years + everyone else with dying MBP's definitly does point to that. Thanks very much though for the insight into the manufactoring/design flaws with this though, very insightful.

Message was edited by: spagbol23