Let's see your best disk speeds( PPC )(68k)

[DarthNvader]

Disk speed is hugely important to responsiveness of the OS and Apps.

So Let's see your best disk speeds for PPC and 68k!

Lately I've been toying with the LSI SAS controllers the have PPC support in Tiger Server and Leopard, but I also have the Sonnet Tempo ATA100 PATA PCI controller and a generic SiL3112 SATA 1.0 controller as well as a few cheap no name IDE to SATA adapters.

I have to test a Beige "G4" a PowerMac G4 QS 800Mhz and a Dual MDD 867MHz along with a PCI-X Dual 2.0 GHz G5.

I'll fill in some benchmarks as I go:

MDD Mac OS 9.2.2 SIL3112 SATA( 33Mhz PCI 32bit ) SSD 92MB/s peak!

Beige G4( 420Mhz/35MHz PCI/70MHz bus )Sonnet ATA100 80GB Seagate spinner Mac OS 9.2.2 52MB/s peak!

Beige G4( 420Mhz/35MHz PCI/70MHz bus )LSI PCI-E x4 in generic PCI bridge Dual SATA SSD's striped RAID( SW ) Tiger Server 10.4.11 87MB/s peak!

 

[chriscarruthers]

What software are you using to produce the benchmark figures?

 

[DarthNvader]

What software are you using to produce the benchmark figures?

I was using the benchmark form Intech HD Speed Tools.

Not getting all scientific, just putting out peak numbers.

G5 Dual 2Ghz Mac OS X Tiger 10.4.11 Sil3112 66Mhz operating in PCI-X slot 2 SATA SSD's striped RAID 161MB/s.( 158MB/s with bus @500MHz and CPU @1Ghz ).

I've got some drives that maybe quicker, but pretty much the Max MB/s we are ever likely to see from a Sil3112.

 

[GorfTheChosen]

G5 Dual 2.7Ghz Mac OS X Tiger 10.4.11, using a pair of 2GB Seagate Barracuda 7200 rpm spinners in a 4TB RAID 0 configuration, housed in an OWC Elite Pro Dual eSATA/USB hardware RAID case, connected via eSATA to a Seritek/2SE4 SATA controller in PCI-X slot 3:

Sequential Uncached Write: 154.77 MB/sec [256K blocks]

Sequential Uncached Read: 151.34 MB/sec [256K blocks]

Tested with Xbench 1.3

Slightly less than half the performance of the 16TB RAID 0 array (around 330MB/s) on the Late 2012 Mac Mini using a somewhat similar case from OWC (but interfaced via Thunderbolt 1)

That OWC eSATA case I'm using on the G5 was a pretty decent value (IMO) ... last time I looked OWC was selling them for less than $60 I believe.

 

[olePigeon]

So who's gonna be the first to use a SATA card with SSDs in a RAID?

 

[GorfTheChosen]

G5 Dual 2.7Ghz Mac OS X Tiger 10.4.11, OWC Electra 3G 240GB SSD, on the G5's built-in (internal) SATA:

Sequential Uncached Write: 121.00 MB/sec [256K blocks]

Sequential Uncached Read: 127.43 MB/sec [256K blocks]

G5 Dual 2.7Ghz Mac OS X Tiger 10.4.11, SPCC 500GB SSD (boot drive/partition), on the G5's built-in (internal) SATA:

Sequential Uncached Write: 116.71 MB/sec [256K blocks]

Sequential Uncached Read: 121.51 MB/sec [256K blocks]

Think that second one is a SATA II drive.

 

[GorfTheChosen]

So who's gonna be the first to use a SATA card with SSDs in a RAID?

I'm sure curious.

 

[herd]

Some of the 64-bit PCI cards can approach the theoretical maximum of 267MB/s in a G4, but I have not found a way to boot from them. With linux I think there is a way to boot on something slow and then change over.

68kmla.org/bb/index.php?threads/g4-raid.40817/

In a G3 (or earlier PPC?), I would be surprised to see anything over 100MB/s because not even the RAM is that fast.

 

[DarthNvader]

G5 Dual 2.7Ghz Mac OS X Tiger 10.4.11, using a pair of 2GB Seagate Barracuda 7200 rpm spinners in a 4TB RAID 0 configuration, housed in an OWC Elite Pro Dual eSATA/USB hardware RAID case, connected via eSATA to a Seritek/2SE4 SATA controller in PCI-X slot 3:

Sequential Uncached Write: 154.77 MB/sec [256K blocks]

Sequential Uncached Read: 151.34 MB/sec [256K blocks]

Tested with Xbench 1.3

Slightly less than half the performance of the 16TB RAID 0 array (around 330MB/s) on the Late 2012 Mac Mini using a somewhat similar case from OWC (but interfaced via Thunderbolt 1)

That OWC eSATA case I'm using on the G5 was a pretty decent value (IMO) ... last time I looked OWC was selling them for less than $60 I believe.

256k blocks is going to be about your slowest speeds, no?

You got me jealous, I've been wanting one of those 2.7GHz AGP/PCI-X Macs for awhile and never can find one close at the right price.

 

[DarthNvader]

Some of the 64-bit PCI cards can approach the theoretical maximum of 267MB/s in a G4, but I have not found a way to boot from them. With linux I think there is a way to boot on something slow and then change over.

68kmla.org/bb/index.php?threads/g4-raid.40817/

In a G3 (or earlier PPC?), I would be surprised to see anything over 100MB/s because not even the RAM is that fast.

The Ram of the Beige starts at 266MB/s, then we start cranking up the bus.

I'm woking on a 66MHz PCI hack for the Beige G3, but I also have a Yikes LoBo I've not touched yet.

 

[DarthNvader]

G5 Dual 2.0GHz LSI SAS3041X two SSD striped RAID Leopard 10.5.8 509MB/s( while spotlight indexes the Mac ).

BTW, the Raid volume is the boot volume.

I still got two SAS ports to stick drives on. 133Mhz PCI-X card.

I should be able to get real close to the 1066MB/s with a full four drive striped array, and it will be bootable!

 

[GorfTheChosen]

256k blocks is going to be about your slowest speeds, no?

Nope ... just the largest size Xbench will test with apparently:



Would have rather used a larger size ... not that it would probably make a ton difference. But who knows ?

You got me jealous, I've been wanting one of those 2.7GHz AGP/PCI-X Macs for awhile and never can find one close at the right price.

The bus speed was a big part of the appeal ... :



The bus speed is being misreported in the above images ... it's actually 1.35 Ghz.

Still would like to have a Quad 2.5 though.

 

[nathall]

Here’s some REAL numbers from a SCREAMER….

Machine is a PowerMac 8500/400Mhz G3. Had to go find a version of Xbench that runs on 10.2.8.

Testing the two spinning disc HDs that live inside. Oh well, it was a fun exercise.

 

[DarthNvader]

Beige G4( 420Mhz/35MHz PCI/70MHz bus )SiL3112 2 SSD striped RAID( SW ) Tiger Server 10.4.11 59MB/s

That's not too bad, better than I thought the SiL3112 would do in the Beige, but it's not 87MB/s either.........

We should get sATAman over here, sometimes he hangs out at os9lives or MacRumors. I know for a fact he worked on a lot of the drivers and firmwares for most of the PCI drive controllers for PPC Macs.

He could most definitely write an 'ndrv' for the LSI controllers that already have FCode ROM's and are bootable in PPC Macs.

With an 'ndrv' we could have working drives in OS 9 and earlier, and still have support in Tiger and Leopard too.

 

[joevt]

Here’s some REAL numbers from a SCREAMER….

Machine is a PowerMac 8500/400Mhz G3. Had to go find a version of Xbench that runs on 10.2.8.

Testing the two spinning disc HDs that live inside. Oh well, it was a fun exercise.

I would like to see the Memory Tests in that screenshot.

 

[GorfTheChosen]

Mac Mini, Late 2012 to the 2 drive 16TB array in that OWC case configured as RAID O via Thunderbolt:

 

[herd]

I would definitely be interested in some sort of fast and bootable storage for a G4 machine. SATA, SAS, nvme, etc... any of these should be able to max out 64-bit PCI.

The Ram of the Beige starts at 266MB/s, then we start cranking up the bus.

That's its theoretical speed, or can you measure it somehow? Every measure of G3 memory throughput that I've seen was unusually low. I don't know why. For example, here is xbench on a Pismo before/after swapping the CPU chip. Note how different the memory speed is, with no other change:

 

[DarthNvader]

I would definitely be interested in some sort of fast and bootable storage for a G4 machine. SATA, SAS, nvme, etc... any of these should be able to max out 64-bit PCI.



That's its theoretical speed, or can you measure it somehow? Every measure of G3 memory throughput that I've seen was unusually low. I don't know why. For example, here is xbench on a Pismo before/after swapping the CPU chip. Note how different the memory speed is, with no other change:

View attachment 82923

What CPU did you start with in the Pismo( 400HHz G3? ) and what did you replace it with CPU? MHz? L2 Cache?

 

[herd]

A 64-bit bus at 100MHz has a theoretical peak speed of 800MB/s (8 bytes at 100MHz). So I'm guessing the 266MB/s that you mentioned is the theoretical peak speed of a 33MHz bus?

Every measure of G3 memory throughput that I've seen was unusually low.

Let me give you another example. I have a 1.1GHz 750G G3 CPU in an AGP with a 100MHz bus (theoretical speed of 800MB/s). Using the AJA System Test with file system cache enabled basically results in a RAM test until the file size exceeds the available RAM. Even with the super G3, the throughput is around 100MB/s. Swapping in a 300MHz 7400 G4 CPU (no other change), results in 350MB/s.

This is why I think that even with a fast 64-bit PCI card for disk IO (267MB/s peak) a G3 machine will not be able to move data any faster than what it can do with RAM.

 

[Powerbase]

A 64-bit bus at 100MHz has a theoretical peak speed of 800MB/s (8 bytes at 100MHz). So I'm guessing the 266MB/s that you mentioned is the theoretical peak speed of a 33MHz bus?



Let me give you another example. I have a 1.1GHz 750G G3 CPU in an AGP with a 100MHz bus (theoretical speed of 800MB/s). Using the AJA System Test with file system cache enabled basically results in a RAM test until the file size exceeds the available RAM. Even with the super G3, the throughput is around 100MB/s. Swapping in a 300MHz 7400 G4 CPU (no other change), results in 350MB/s.

This is why I think that even with a fast 64-bit PCI card for disk IO (267MB/s peak) a G3 machine will not be able to move data any faster than what it can do with RAM.

Yeah the G4s supported the newer/better MPX bus while the G3s were stuck on the older 60x bus.