1.25Ghz eMac overclock

[MacTron]

Now we have the 1.25Ghz eMac working really well with Mac Os 9, is time to extract more juice to this surprisingly fast machine, it's time to try the overclock

Here have the main info (in French):
https://sites.google.com/site/overclockmac/g4emac2

Usually the 1.25 eMac can be overclocked up to 1.5Ghz without increasing the CPU voltage. So only 3 resistors of the motherboard must be removed for this easy overcklock:
R270 (front side), R673 (reverse side) and R689 (reverse side).




There is reports of the  1.25 eMac working fully estable at 1.75 Ghz with the core voltage upped to 1.42:



I will post detailed pictures about were this components are located.

[MacTron]

This first (and basic) test show us that the 1.5 Ghz eMac is faster than my MDD 1.5Ghz with 2Mb of L3 cache! . Unbelievable:


The extra speed in the Disk Mix of the 1.5 is also due to an HD replacement. I have changed the 40Gb  stock  HD for a faster WD 320Gb
The  superiority of the MDD in the video test is due to the unbeaten nVidia ti4600 video card.

I will check the system stability in the next few days ...

[darthnVader]

Quake III timedemo, or it didn't happen.

[macStuff]

did u actually do the overclock to 1.75ghz? 

[darthnVader]

It's interesting that the eMac had a 167Mhz Bus, while the iBook G4 had a 133Mhz Bus, and both used PC2700 Ram.

As we'll recall the G4 never supported DDR, meaning we never got to take advantage of the Double Data Rate of the Ram. So the Ram just runs at the bus speed.

Finding the PLL that sets the Bus, we should be able to increase the performance quite a bit for tasks the are memory bandwidth bound, at least for the 133 Mhz bus models.

Also, Macbench isn't really the best benchmark, it was never pure carbon, so you don't know how much of the 68k emulator you are benchmarking vs. PPC code. Plus I don't think they ever updated it to take into account CPU Cache.

Real world benchmarks bear out the 7455 with it's L3 cache is much faster, even at a lower clock speed than any 7447 could be. In particular 3D gaming, however if the task can't take advantage  of the L3 cache clock speed wins out, and the 7447 maybe faster clock for clock than the 7455.

http://www.xlr8yourmac.com/G4CARDS/powerlogix_dual_g4_7447_7457/index3G.html

[MacTron]

It's interesting that the eMac had a 167Mhz Bus, while the iBook G4 had a 133Mhz Bus, and both used PC2700 Ram.

... it depends on the exact model. There is different bus speeds and  different RAM speeds modules -consequently- for both computers.

As we'll recall the G4 never supported DDR, meaning we never got to take advantage of the Double Data Rate of the Ram. So the Ram just runs at the bus speed.

... but the memory controller support DDR, in fact this is really useful for most DMA process that travel along the PCI and AGP busses.

Also, Macbench isn't really the best benchmark, it was never pure carbon, so you don't know how much of the 68k emulator you are benchmarking vs. PPC code.

I disagree. MacBench runs exactly the same as 95% of Mac apps. Carbon apps are really scarce and dam slow.
I really prefer CineBench for deep and acurate testing.

Plus I don't think they ever updated it to take into account CPU Cache.
Real world benchmarks bear out the 7455 with it's L3 cache is much faster, even at a lower clock speed than any 7447 could be.

Yes, that's it. I the 7447 at 1.83Ghz performs similar to a 7455 at 1.66Ghz in the MDD by example. I have tested it.
But the results in the eMac break this ... I don't know why, may be the Intrepid IC ¿?

[MacTron]

did u actually do the overclock to 1.75ghz? 

No i didn't. ( by now  ). But people here: (in french):
http://forum.macbidouille.com/index.php?showtopic=255082
I still have to fully test the stability of the 1.5 overclock.

[MacTron]

The 1.5 overclock runs fully stable by now, after a couple of video encoding test, the last one lasted over two hours without any issue.
Those are the Cinebench 2003 test results compared to a 1.5Ghz MDD with 2Mb of L3 cache:



The 7447 in the eMac performs better than espected.

[macStuff]

too bad you never made any youtube video so people can see MARL labs in action

[MacTron]

too bad you never made any youtube video so people can see MARL labs in action

It take a bout an hour to disassembly and reassembly the eMac, to record and watch a video of it, will be really boring...
The overclock itself is easy and fast.

[DieHard]

I am guessing the eMac runs much quieter even when rendering ?

[MacTron]

I am guessing the eMac runs much quieter even when rendering ?

Yes, that's it. The 7447 is 1/2 or 1/3 colder than a 7455 at the same speed, IIRC.
The eMac isn't as quiet as my iMac G4, but definitely is quieter than my MDD. May be it produce half the noise ( or even less ) than my MDD...

[IIO]

guys take care when disassembling units with CRT monitor.

[macStuff]

ive always loved the power but detested the form factor + weight of the MDD machines
thats why the digital audio / quicksilver was my favourite and how come i still will mess around with a b+W g3 or a sawtooth
but i havent tried a mac mini running os 9 yet id love to get a digi 002 console teamed up with a mac mini running 9.22 - thats heavy duty + travelling light at the same time; i bet the room would be nice + quiet

[reader50]

Could someone fix the thread title, please?

Sorry, I have nothing to offer regarding an eMac. But the MHz part catches my eye every time this thread is bumped. 1.25 MHz takes me back to the Apple //e days, with an accelerator card.

[MacOS Plus]

  Good catch!  Otherwise we'll have to change the title to "underclock". (Which was also typo'ed as "overcklock" anyway!)

[MacTron]

OK, fixed typos .    Thanx