Amiga.org
Amiga computer related discussion => Amiga Hardware Issues and discussion => Topic started by: HardStep on May 02, 2006, 08:48:20 PM
-
Hi guys,
i was considering overclocking my Blizzard1260 from 50mhz to 66mhz. Before i move on with replacing the crystals, i´d like to hear about issues with longtime reliability and about overall speed gains.
When using one inside desktop case, is there a possibility to use some ultraslim fan on 060 without lifting the machine a bit?
-
In the days when we learned our stuff from the likes of Amiga Format and CUAmiga, we were advised to limit overclocking any 68-series CPU to 15% of its stated speed. That means any standard 50MHz chip shouldn't be pushed beyond 58MHz or so. Having said that, plenty of people have upped their 060 to 66MHz and beyond. I've read of them running at 80MHz or so, but how long they lasted is another issue. Running a CPU above its design spec will reduce its lifespan, but individual chips vary.
I run my Apollo 1260 at 64MHz (the fastest oscillator Maplins sold) and never had a problem with it in ten years. It is however, in a tower case which will make a difference. The standard A1200 case is pretty restricted and though the 060s run relatively cool, upping it to 66MHz will introduce a need for more air. I once fitted a standard 030/50 in my desktop and that would heat crash when running Imagine renders. Jacking up the case and leaving the trapdoor off solved it. A fan also helped but it was very loud.
To a large extent it may depend on what you use it for. My tower has a heat sink on the CPU but no fan, it wasn't necessary and was noisey. Don't know if an 060/66 runs cooler than an 030/50, but I'm sure someone here will tell us.
My advice is go ahead and have fun.
JaX
-
I know several Atari users are running 060's overclocked to 100Mhz in their Falcon but they seem to have very big fans (http://www.czuba-tech.com/ct60/CT60_inc.jpg) installed. I haven't heard of anyone going over 80mhz in an Amiga.
If you remove the crystal and place a socket there you can experiment with different speed crystals to find which speed works best and most stable. Let it run 68k quake for an hour or so to test.
On the AIBB benchmarks page (http://amiga.resource.cx/perf/aibbde.html), an 060/50Mhz gets 26.74 Dhrystones, and 060/64Mhz gets 36.26, which is in line with the clock speed increase.
-
HardStep wrote:
Hi guys,
i was considering overclocking my Blizzard1260 from 50mhz to 66mhz. Before i move on with replacing the crystals, i´d like to hear about issues with longtime reliability and about overall speed gains.
When using one inside desktop case, is there a possibility to use some ultraslim fan on 060 without lifting the machine a bit?
Had my Blizzard running at 66MHz for 7years before the machine and I parted company.
I had a small heatsink attached with a metal clip around the accellerator board (insulated) plus a thin smear of heatsink grease between it and the processor.
Even at 66MHz it never got as hot as the 030/50 on one of the other machines.
Tried 75MHz at one point but found some of the FPU instructions became unstable. Chip appeared to work fine as long as no FPU instructions were used.
Another side effect at 75MHz was that the SCSI board stopped working.
Preferred to stick with 66Mhz, and a stable Amiga.
Cheers,
Red
-
I know several Atari users are running 060's overclocked to 100Mhz in their Falcon but they seem to have very big fans installed. I haven't heard of anyone going over 80mhz in an Amiga.
this fan is like the ones on 486 cpus... not a very big one :lol:
-
I believe the Atari users are using a very late mask that allows the faster clock speeds. I wondered if those chips could be clocked to those speeds on the amiga. I have a MKII with a socketted oscillator so if I can get one of them I could try but I think it will be a limitation of the board and 32-pin SIMM speed. They are using SD-RAM chips I believe which are faster.
-
Didn't some of the Atari Falcons have a faster rated 060 to start with?
-
Hi,
I've had my phase 5 Blizzard 1260 running at 66MHz for oh, must be 7 or 8 years now. Still no problems, and although my machine isn't very stable to begin with, extensive testing has shown that overclocking doesn't make it any more unstable. Also using the SCSI kit with it, but haven't had the opportunity (or motivation) to see if its speed was improved in any way by the new crystal. I did put a socket on the board at the time though so that I could swap back should any problems start to appear,but it runs cool in my tower anyway with a small 486 fan on it... I suppose 486 fans are getting hard to come by at this stage, so a chipset fan would do the trick instead.
-
The reason that the Falcon accelerator (Centurbo 6x) can operate at upto 100MHz is 2 fold.
1) They use a much later version of the MC68060RC50 (E41J) which runs MUCH cooler than the earlier ones
2) The Centurbo has 133MHz SDRAM interface.
Running much beyond the 66MHz range is not possible with an amiga accelerator due to the fact they use SIMMs
-
@ alexh,
Not entiry true you can get 060 running at 75MHz from Apollo and these still use SIMMs. Although they are not full 060s (they lack an FPU)
-
exactly the fpu speed is the prob on the 50mgz 060 and that's why they run a cut version.
-
It could run fast if you would use 50 ns Simms or even faster if they exist. But that would rule out the possibility to use 128 Mb Simms as they are 60 ns :boohoo:
-
The Atari falcon only came with a 030 @ 16MHz stock. A Falcon 040 was in the works (protos were made) but Atari folded at that point.
-
Back when I was messing around with 486 PCs, I found by experimentation some SIMMs (although they were labeled as 60ns) that could reliably give me a cache line in 18 cycles on a 50MHz bus. I think that is almost what the performance of a 40ns SIMM should be.
-
but what could this in practice mean for an overclocked 060?
i have a crystall at 66 and 2x32mb simms at 50ns but can this be safe and stable? more than this can be safe at higher speeds with these simms?
-
@AmiBoy
The 75MHz version of the Apollo is probably designed to use more cycles when accessing RAM.
Classically the memory controller on a 50MHz 060 board appears to use 4(?) cycles on RAM access. The 75MHz "firmware" I speculate uses more. I could be wrong however (13ns * 4 = 52ns which may be within the capabilities of SIMMS).
-
these more ''wait states'' finaly result in a less
mips performance/mhz?
-
To be honest I dont quite have the knowledge of the relationship between 680x0 CPU cycles (i.e. 50MHz 20ns) and 680x0 Bus cycles, especially when you start to include RAM pre-charge & burst access not to mention synchronous and asynchronous access.
If you look at some of the performance figures of 50Mbyte/s that most 50MHz Amiga 060 cards had it equates to 4 cycles per access. (4 cycles = 80ns)
If you look at the performance figures for the Centurbo they also average 4 cycles per access. However 4 cycles at 100MHz is 40ns, beyond the capability of SIMMS.
-
you can run apollos and blizzards to 80 mhz and above i have overclocked apollo 1260 to100 mhz blizzard to 80