Separate names with a comma.
Discussion in 'PowerPC Macs' started by Dronecatcher, Jun 13, 2016.
EDIT: Companion benchmark test
Honestly, I wish I'd been into PPC Macs when Daystar was still doing this. I'd send one in-granted I know it's no longer financially practical, but I've wasted money on worse things.
I have the same processor-a 2ghz 7448-in a tower and it's amazingly fast. You get 1mb of full speed on-die L2, which perks these up an amazing amount. Mine is fast in a Digital Audio with it 133mhz FSB...I can only imagine what the much faster DLSD architecture would do.
Even more importantly in a laptop, the 7448 actually draws LESS power than the 7447A. The fastest 7447As I have are 1.8ghz(actually I think the Sonnets are 1.42s overclocked to 1.8) and they are HOT. The 7448 runs noticeably cooler. Of course, in a laptop this should also translate into better battery life.
I was literally just Googling overclocking PowerBook G4s to 2GHz. Spooky.
I'd give this a go, but my PowerBook is my baby and the thought of overclocking it both excites me and terrifies me to the core. I remember when Daystar were offering 1.8/1.9/2.0GHz overclock upgrades around 2006, and there were just tonnes of horror stories of how it made them run awfully slow, super hot all the time, burn holes through tables/people's laps or just plain break and need a new logic board entirely.
Wow, I wonder, is there anywhere I can get one with the Daystar upgrade? I checked all over eBay, and could not find one.
They weren't overly popular IIRC, besides, all they did was overclock the processor.
Well then, is there any way for me to do that myself? I thought it was impossible to overclock a PowerBook, you can fake the frequency via software or the OpenFirmware, but that does not help performance any.
Here's a thread that has a lot of details about how to overclock various Macs (seemingly concentrating mainly on PowerBooks) by moving jumpers.
Daystar completely replaced the processor with a 7448 vs. the 7447A that shipped in the DLSD(and most other AlBooks). It's not just simply overclocking it, and the swap is significant as the TDP of the 7448 is-as I recall-about half that of the 7447A.
The computer had to go to them for the upgrade.
Doesn't the DLSD PowerBook have a 7448? I was pretty sure it does... And I was only thinking of the Daystar upgrades for DLSD PowerBook, opposed to earlier 7447a based AlBooks. And I'm pretty sure the 7448 processor can be overclocked by moving a few jumpers. I remember a friend I had claiming he had overclocked his PowerBook using that method up to 2.8GHz!
No, no Apple computer shipped with a 7448. They were only available as aftermarket upgrades.
I'm finding a lot of conflicting information, EveryMac says it "actually has a PowerPC 7448 (G4) processor as required to interface with the PC2-4200 DDR SDRAM"; various sources online say both 7447 and 7448, and MacTracker says it has a 7447b (this sounds most likely).
Edit: Did a bit more research, it is a variant of the 7447, as Bunn and Intell said. Don't know why I was thinking it was 7448, wishful thinking, perhaps.
Edit 2: I've been having a bit of a think, and remembered a possible reason why I thought they were 7448, I was getting mixed up with a rumour from 2005 that the DLSDs were originally supposed to have the 7448, but Apple pulled out at the last minute. And following the release of 7447 DLSDs, there were rumours that there would be a final 'PPC send-off' PowerBook featuring a 7448 to be released in Q1 2006.
I've seen the die markings on the DLSD Powerbook, they are a 7447 variant.
I did a bit more digging and found the OP of the video over in the 68kmla forums, and he states:
I wish this myth about the DLSDs having a 7448 would die a slow and painful death.
Not that I didn't believe Intell, but as it so happened I had an already halfway taken apart A1138 easily accessible so I pulled the logic board a few minutes ago. It's marked 7447B plain as day on the CPU die. Unless this is another Rabidz7 myth about them "hiding" features, that's what it is.
I struggled with actually getting a legible die shot-I'll try again this evening. In any case, here are photos. The first is the complete logic board to show the DDR2 RAM(which, by the way, doesn't "require" any certain kind of processor) and the second is the most legible die photo I could get in the few minutes before I left for work. You can at least also see the 7447 marking on the processor.
In any case, the litmus test for a 7448(there's no 7448A, by the way) is the size of the L2 cache. If the cache isn't 1mb, it's not a 7448. I fire up my A1139 when I get to work, and I will fire up a real, known 7448 this evening to show this.
Thank you very much for that!
I wonder where this 7448 myth originates from then! I've heard it more than once, including the "7447A with DDR2 RAM is impossible" line. And why does OP insist so adamantly that it's a 7448, even though he no doubt will have seen the die himself...
Edit: from OP's Geekbench results:
I suppose OP must be lucky enough to have a A1139 with a (possibly Daystar) aftermarket processor upgrade.
Also, IIRC, isn't 1.7GHz the highest a 7447A can be overclocked to?
This has rabidz7 written all over it
Like I said, I have a 7448 and will show what it looks like in sys. profiler and Geekbench this evening.
Here's the system profiler shot of the A1139. Note the cache size.
Relevant Open Firmware information
Question: Are G4 CPU upgrades plug and play, instantly identified by the OS and correctly reported by System Profiler or along with the hardware is there a software/driver installation?
It largely depends on the upgrade.
745x upgrades in 7400 systems are plug and play and will generally report correctly in OS X.
The 7447A based upgrades(which are fairly common) require a firmware patch on G4 towers to boot and depending on the upgrade sometimes spoof another processor.
In any case, as I said earlier the cache size is a giveaway. If it has L3, the processor is a 745x. If it has 1mb of L2, it's a 7448. The L3 thing is iffy as a lot of 745x processors didn't have it, and I've also found a quirk in Quicksilvers where the L3 doesn't generally show up if the computer didn't originally ship with it. As far as I know, though, the 1mb L2 is an absolute positive/negative test for a 7448. 1mb L2=7448, less than 1mb=not 7448
--- Post Merged, Jun 16, 2016 ---
Sonnet reliably got them to 1.8ghz. I have them in both single and dual configurations. I have a dual Giga that claims to be able to do 1.8, but neither the previous owner nor I have been able to get it there. I have it ticking along happily now at 1.6, although the previous owner did have it running at 1.73 for a while.
--- Post Merged, Jun 16, 2016 ---
My reasoning was the possibility of having an upgraded Powerbook that isn't acknowledged by System Profiler correctly - hence the L2 appearing as normal.
I can't find a way of reading back the processor ID on my Powerbook - in Terminal, "machine" gives 7450 and "system_profiler SPHardwareDataType" gives PowerPc G4 (1.5) - I think profilers just read back a text string (hence people spoofing systems).
Having watched the video and read the guy's forum piece he doesn't come across as someone who'd orchestrate a lengthy, pointless hoax.
Makes you think though.....my 17" benchmarks correctly but it's always felt smoother and faster than it rightly should.
Here are a few relevant photos from the 7448 computer I have. This is a 2ghz Newertech upgrade. System profiler reports the processor as 60? , but again the give-away is the 1mb of L2.
I'm also posting a photo from the graphics tab, as this is the only X800(actually a flashed FireGl X3) that I've seen working in a Digital Audio or Quicksilver G4.
I've asked the owner to post here - he appears to have a lot of technical detail on this machine - anyhow, he said there'd be more videos in the pipeline to illustrate the performance, so we can make our own judgements from that.
Afternoon all, me releasing this thread/video has seemed to really kick the hornets nest more than I had anticipated. I noticed a lot of you think I'm playing some ruse here with a 7447A, Which I can totally understand but this machine is indeed a 7448 based chip chugging away at 2Ghz on the stock 1.3v. There is no way you'd be sitting cinebench stable at 1.3v on a 7447, but regardless I do indeed have factual proof this is a 7448. Personally I really don't want to tear this thing down to ribbons to take a die shot, but if I must that can be allocated when i find the time.
In addition I do have Apples very own Internal Engineering documentation (Both revisions) (And board designs for manufacturing) regarding the story I posted about. But I don't know if posting them online is going to get me into any kind of Intellectual Property problems.
As of right now I'm trying to get the machine back up after screwing with the FSB to run it back at 200Mhz. Turns out that was a bad idea as their is a bit to be set that makes the FSB/PLL chips take in the configuration table as read only. So now I've to figure out how to set that back to stock. None the less I'm more than happy to provide any details that may be needed to prove this is indeed the truth.
I don't mean to be a rear end about this, but I've posted die shots and Open Firmware profiles from DLSDs.
I'm going to be blunt and say that the only thing at this point that would convince me is a die shot. If the die says 7448 on it, I will congratulate you on owning an excellent piece. Otherwise, I remain sceptical. I've put some time into substantiating my position with real-world examples.
I was watching your video when I noticed at the 52 second mark you mention that the "7448" has a "platform controller hub" to allow DDR2 ram to work. I'd like to bring to the attention of readers that in a G4 system the CPU in no way handles or has anything to do with memory support or controlling memory directly. A good example of this is a 7455 sonnet upgrade for say a 9500 which was made in 1995 and uses EDO ram yet the same CPU can be found in G4s with DDR1 ram. In a PPC Mac the thing that handles memory support is eather a dedicated controller in the 9500's case it's the hammer head controller, in later systems the memory controller was integrated in the north bridge and in the DLSD this is known as intrepid 2 following intrepid in the earlier ally PowerBook G4s, note in the DLSD block diagram the CPU is on its own MaxBUS a bus every G4 has had since the Sawtooth. Here are some sources to back up these claims
I'm very much aware of that, sorry If I had made a small mistake. I mean to say the Second generation Intreped 2 Platform Controller allows DDR2 work, which featured a DDR2 memory controller. I will be remaking the videos, and be starting a series.
I had posted pictures of the die shots, motherboard and schematics for a few hours on here (I'm sure many of you saw them), but I've ultimately decided to pull them down and instead do a more in depth review/analysis of the machine and answer some of the more repetitive questions. I've gotten far more attention/inquiries than I'd initially anticipated, and figure it deserves a proper quality content series rather than a bunch of vague posts and comments on forums.