Mac OS 9 Lives
Mac OS 9 Discussion => Hardware => CPU Upgrades => Topic started by: InspectorG on March 02, 2017, 12:06:58 AM
-
I have the opportunity to get a 2x 1.25GHz card from a 2003 FW800 MDD for cheap, and I'm curious if this will work in my 2003 FW400 MDD. I've tried searching a bit but can't find a definitive answer. Anyone know the details in this regard?
And just for the sake of posterity, what about a 2x 1.42GHz card? Perhaps someone else in the future will want to know this too.
-
It will work just fine, but depending on what specs your MDD has it might not run at full speed (the 866mhz MDD bus is clocked at 133mhz compared to the 167mhz bus the Dual 1.25ghz version uses).
-
We are, of course, assuming that since you're considering a 2x1.25 as an upgrade, your MDD must have a single proc and therefore is clocked at 133mHz.
**All MDD CPU daughtercards are a drop-in fit on any MDD motherboard.
The clock speed can be easily changed from 133 to 167mHz by the removal of a resistor. The procedure can be found by Googling "MDD clock speed" (duh). Note the only caveat to this: It's fairly easy to remove the SMD resistor but extremely difficult to replace it. If your "new" CPU turns out to be defective and you have to replace the single, you'll have a crash-every-5-minutes computer (if it runs at all) without slowing the clock back down. So…
Install the 2x1.25 CPU and run it at 133mHz for a couple of days to be certain it's good before you amputate!!
AND…
You absolutely must do a good job with Arctic Silver or other good heat sink compound when you change over. The dual will obviously generate more heat than the single so the fan may (who am I kidding…WILL) run harder. Although the "thick" aluminum sink will work fine, if you have the "thin" finned aluminum or have the opportunity to get one or even a copper finned sink from a 2x1.42, do it.
-
Thanks for the detailed info guys!
I do have a single 1.25 GHz 2003 FW400 MDD, but the bus is clocked at 167MHz so I'm good in that department. However, I am not sure if this model has the "thick" or "thin" fin aluminum heatsink. Is it as simple as looking at it to determine "thick" or "thin" and comparing to some pics? Anyone have comparison pics handy?
-
Is it shiny sheets of silver metal (thin) or 1-piece cast aluminum (thick)?
Either way, you shouldn't have any serious issues. Just get the good heat sink compound.
-
Is it shiny sheets of silver metal (thin) or 1-piece cast aluminum (thick)?
Either way, you shouldn't have any serious issues. Just get the good heat sink compound.
Thanks for the clarification. It appears to be of 1-piece construction, so is presumably the thick-fin version. I might try to get my hands on one of the better heatsinks in the near future then.
Has there been much experimentation in aftermarket heatsinks? I've seen a few pictures floating around but no definitive/organized information.
-
There's been no real mucking around like that as far as I know - probably because the originals have been fairly common and work well enough. Don't lose sleep over it… just snatch one if one happens to come by.
-
Is this the thin fin heatsink in question? I'm having a little trouble finding the right part number, as many of the sites publishing Apple's part numbers are giving conflicting information.
http://www.ebay.com/itm/371841178497
-
The Dual 1ghz and Dual 1.25ghz MDDs came with the "thick" aluminium heatsink, the 867mhz MDD with the thin Steel heat sink and the dual 1.42ghz came with the copper heat sink.
Regarding thermal performance the thin steel is the worst, thick aluminium better and copper best. Have a look at this for comparison: http://macos9lives.com/smforum/index.php/topic,915.msg3266.html#msg3266
I myself have been using my dual 1.5ghz (overclocked from 1.25ghz) CPU with the thick aluminium heatsink for 7-8 years or so before replacing it with the Sonnet MDX I have in it now. No issues to report what so ever. So if you really want something better, get a copper heatsink from a dual 1.42ghz MDD, but you don't have to :)
-
Yeah you don't have to, but that cooper one looks awesome!! Haha! ;D ;D
-
Regarding thermal performance the thin steel is the worst, thick aluminium better and copper best. Have a look at this for comparison: http://macos9lives.com/smforum/index.php/topic,915.msg3266.html#msg3266
That thread is amazing! Round and round and round and at the end there's still disagreement over the performance of the three types. That is exactly why I said "don't lose any sleep over it".
My opinion of the relative performance of the three is completely subjective but it does take into account my impressions of both actual thermal efficiency and the "side effects" of overall noise levels caused by both airflow noise related to the different shapes and the amount of fan RPM (and therefore noise) my computer apparently felt necessary to maintain proper temps.
There are sooo many variables - fan make/model, CPU load, air restriction caused by more HDDs and/or PCI cards, more ambient heat caused by more HDDs and/or PCI cards, ambient room temp etc. etc. - that accurately comparing the three is guaranteed to have one reaching for the Tylenol very quickly.
At the end of the day, any of the three will more or less work with any of the CPU cards. So, I again repeat my fully researched scientific analysis:
"Don't lose any sleep over it"
-
The Dual 1ghz and Dual 1.25ghz MDDs came with the "thick" aluminium heatsink, the 867mhz MDD with the thin Steel heat sink and the dual 1.42ghz came with the copper heat sink.
My 2 MDD dual 867 have "thick" aluminium. And they both came with the original "wind tunnel" fans/PSU.
The "thin" ones came with the higher speed ones IIRC.
The thin one is supposed to be "better", but they "age" in a worst way than the thick aluminum ones. I mean, the heatpipes "get older" more quickly than the aluminium one and could have less performance (in case the tubes loose the gas) while the thick one has no moving parts
-
Nor the thin or the thick have heatpipes, only the copper one. My mistake.
But the thin one can have worst performance if the fins are "loose". If they are "tight" should be better.
MacTron in his M.A.R.L. lab had better performance with the thick ones IIRC.
-
The "thin" ones came with the higher speed ones IIRC.
Strange. My old studio MDD started life as a Dual 867mhz and it had the thin heatsink. My current home MDD started life as a Dual 1ghz MDD and had the thicker aluminium heatsink.
It might have been more like They took whatever was on the shelfs? Still, i have had no issues with the thick heatsink. Whatever you have will work just fine :)
-
he is talking about a 2003 fw-400 ... so that doesnt affect 833 or 875 mhz ... there was only a dual 1,25.
i remember it well, apple sold it for almost one year longe rthan the fw-800 model, and it was far more expensive than the fw-800. :)
-
(http://macos9lives.com/smforum/index.php?action=dlattach;topic=587.0;attach=574)
-
i stand corrected (images are always right); there was only a single 1,25.
-
I swapped in the dual 1.25 GHz card today, and kept using the thick fin heatsink I already had. However, I also bought a thin fin heatsink for cheap for comparison. The part number 805-4759-A is indeed the correct item.
The thick fin HS is far more substantial; it weighs 1.4 kg, whereas the thin fin HS weighs only 0.8 kg. This is more mass to absorb heat, but also more mass from which heat needs to be removed. I think the additional mass will ultimately prove to make the thick fin HS better overall though.
I see from that table of MDD G4 chips that the FW800 has 7455A chips, whereas the 2003 FW400 has 7455B chips. Are there any differences between the two worth noting?
-
I think the additional mass will ultimately prove to make the thick fin HS better overall though.
…and right there you're off into the woods with just how complex something that appears simple can actually be.
1. Mass in and of itself is not helpful with thermal efficiency - what counts is thermal conductivity - how fast heat will move through it.
2. The "thin" sink assembly, while it may appear to be less "substantial" than the heavy cast one, actually has more surface area to pass the heat off into the air.
3. The biggest "problem" area on all of these is the immediate vicinity of the CPU's. That's where the heat is and what counts is how fast it gets carried away from the chips before spreading out into the sink and finally dissipated into the air. This is what makes the copper model superior with its gas tube. It's very good at preventing a "hot spot" right above the chips.
Like I said, they all do a pretty good job and while we can endlessly debate thermal dynamics, at the end of the day it's all academic.
-
The aluminium thick one is better than the thin one, because in the thin one the thin plates (made of zinc IMHO ) are soldered to the main aluminium base, so this leaves to a less efficient heat dissipation.
I have posted the data somewhere in this forum ...
-
…and around and around we go…
MacTron, I have the utmost respect for you but I think you are missing my point.
Although you are probably correct that the "thin" plates are zinc, They are not soldered but rather "welded" to the base plate by some sort of (MIG?) process that is not affected by whatever temps the base plate reaches and they stay put. Notice how each one is bent 90° to maximize the contact area. The heat isn't flowing through a solder paste or such.
BUT
I am not debating only the absolute thermal performance of the 3 types. I really don't think that any of us here on the Forum are equipped with sophisticated equipment to accurately compare their real-world thermal performance. This also requires some elaborate advanced math (I am not ashamed to admit is way beyond me) just to compare say, the radiating efficiency of the 24 thick fins on the one versus the 39 thin ones on the other. However, since all but the two outer ones on each radiate into the narrow spaces between themselves and their neighbors, the actual airflow in those spaces, both its incoming temperature and its velocity become a huge factor. Therefore, changing the stock fan out for any other unit will affect performance regardless of the fan noise - for example, two virtually identical fans with identical "rated" airflow but with slightly different blade shapes may create slightly different airflow vortexes that cause significantly different volumes of air to actually flow through the heatsink.
So, you see that I am trying take into account the practical overall performance of them in actual use.
A ridiculously simple test: Hold up the two "silver" sinks in front of you. Blow through them. You will find that while the "thin" unit may rattle slightly, it's generally pretty silent. The "thick" cast one however, will immediately resonate and ring quite loudly. They do this all day long! This noise is added to the air handling noise at what would otherwise probably be a relatively quiet frequency. This is just one of the many noise sources that caused the damn MDD's to be dubbed "The Windtunnel".
So, I reiterate my points:
1. All engineering solutions are a compromise between cost and effectiveness. If there was a simple way to really reduce the MDD noise without compromising reliability, Apple would have done it.
2. There are no "magic fans" that somehow work silently while still providing the same airflow. There are improved models, sure, but if you're going to play engineer and redesign something as critical as computer cooling - especially on a unit that uses a chip design that's being pushed to its absolute limit and runs too hot to begin with, you had better be prepared to survive the occasional catastrophic failure that may result.
3. Everyone's individual experience will be unique. All other factors ie. CPU load, preferred ambient room temp, personal tolerance level, furniture layout etc etc will cause some to be perfectly happy while others are insufferably annoyed.
And finally…
4. Most, definitely most of the noise comes from the damn PSU fans anyway. All of the above applies here as well but even more so. There is little or no margin for error here.
That's it…I'm pooped.
-
"If there was a simple way to really reduce the MDD noise without compromising reliability, Apple would have done it."
:D