Codenamed "Sandbox"? The name I've seen for a VMX enabled G3 is "Mojave" or 750VX.. Sandbox was new to me.
Correct.
If you want to see if it's a G3, test the FPU. A G4 of the same clock speed should have about double the FPU performance of the G3 with VMX. I'm pretty sure the 750VX is being saved for an Apple product yet to come.
If you want to see if it's a G3, test the FPU. A G4 of the same clock speed should have about double the FPU performance of the G3 with VMX.
First.. we know nothing about the 750VX. Second.. the G4 has only about 10% higher performance than the G3 when doing regular FPU work at the same frequency.
Generally, both the G4/74xx (Motorola) and G3/7xx (IBM) CPUs are descended from the old PowerPC 603 CPU, and thus have similar performance clock-for-clock, except for the presence of Altivec on the G4 and differing amounts of L2/L3 cache, etc.
The tool you guys are using only gives processor families. The 7400 and 7410 are essentially the same chip; the 7450, 7451, 7455 and 7457 are essentially the same chip; etc. It's using the name of the first CPU in the family. For that matter, there are even different kinds of 7455. Also, the 7440, 7441, 7445 and 7457 get reported as 7450s, because, for all intents and purposes, they are.
The 745x family have tags for L3 cache, which means that they can support L3 cache. The amount is up to the board maker, although I believe the limit is 1MB on the 7455 and 2MB on the 7457. The 744x versions are smaller (and therefore cooler and cheaper), because they omit the support for L3 cache. Any PowerBook, iBook, eMac or iMac with no cache probably (though not necessarily) has a 7445 or a 7447.
More details are available here, at Mot SPS. Interestingly, the 7457/7447 aren't listed as in production (which means, as I see it, that Apple is getting every single one of them), but if you click on their names you'll get a product summary anyway.
XBench reports the processor with greater accuracy than the Terminal does. My eMac is listed has having a 7455 according to XBench. Therefore, I am not through with this platform and therefore I will not be buying a much faster, cheaper PC that can actually use floppies and a two button mouse.
First.. we know nothing about the 750VX. Second.. the G4 has only about 10% higher performance than the G3 when doing regular FPU work at the same frequency.
Well, from a simplified perspective, the g3 must execute each double precision floating point instruction twice. The g4 does not. So if the code were purely double precision FPU work, then indeed I would presume a big gain on the g4.
Comments
Originally posted by Henriok
Codenamed "Sandbox"? The name I've seen for a VMX enabled G3 is "Mojave" or 750VX.. Sandbox was new to me.
Correct.
If you want to see if it's a G3, test the FPU. A G4 of the same clock speed should have about double the FPU performance of the G3 with VMX. I'm pretty sure the 750VX is being saved for an Apple product yet to come.
Originally posted by Splinemodel
If you want to see if it's a G3, test the FPU. A G4 of the same clock speed should have about double the FPU performance of the G3 with VMX.
First.. we know nothing about the 750VX. Second.. the G4 has only about 10% higher performance than the G3 when doing regular FPU work at the same frequency.
Kernel configured for a single processor only.
1 processor is physically available.
Processor type: ppc7450 (PowerPC 7450)
Processor active: 0
Primary memory available: 256.00 megabytes.
Default processor set: 48 tasks, 166 threads, 1 processors
Load average: 1.66, Mach factor: 0.06
I know that my CPU (G4 7450) has 256kb of full speed L2 cache. No L3.
So what does the 7455 have?
The 745x family have tags for L3 cache, which means that they can support L3 cache. The amount is up to the board maker, although I believe the limit is 1MB on the 7455 and 2MB on the 7457. The 744x versions are smaller (and therefore cooler and cheaper), because they omit the support for L3 cache. Any PowerBook, iBook, eMac or iMac with no cache probably (though not necessarily) has a 7445 or a 7447.
More details are available here, at Mot SPS. Interestingly, the 7457/7447 aren't listed as in production (which means, as I see it, that Apple is getting every single one of them), but if you click on their names you'll get a product summary anyway.
Other systems using the 7455:
iBook G4
PowerBook 12" G4 867 MHz
PowerBook Titanium 800 MHz-1 GHz
iMac 1.0-1.25 GHz
PowerMac G4 single processor 933 MHz-1.25 GHz, dual processor 867 MHz-1.42 GHz
I guess the newer AlBooks have 7447s or 7457s that don't report their information to XBench normally.
Results\t84.03\t
\tSystem Info\t\t
\t\tXbench Version\t\t1.1.3
\t\tSystem Version\t\t10.3 (7B85)
\t\tPhysical RAM\t\t256 MB
\t\tModel\t\tPowerBook6,3
\t\tProcessor\t\tPowerPC G4 @ 934 MHz
\t\t\tVersion\t\t7455 (Apollo) v3.3
\t\t\tL1 Cache\t\t32K (instruction), 32K (data)
\t\t\tL2 Cache\t\t256K @ 934 MHz
\t\t\tBus Frequency\t\t133 MHz
\t\tVideo Card\t\tATY,RV280M9+
\t\tDrive Type\t\tTOSHIBA MK4025GAS
\tCPU Test\t108.79\t
\t\tGCD Loop\t99.65\t3.89 Mops/sec
\t\tFloating Point Basic\t113.92\t411.96 Mflop/sec
\t\tAltiVec Basic\t113.36\t3.29 Gflop/sec
\t\tvecLib FFT\t113.78\t1.77 Gflop/sec
\t\tFloating Point Library\t104.85\t4.20 Mops/sec
\tThread Test\t75.76\t
\t\tComputation\t55.35\t747.22 Kops/sec, 4 threads
\t\tLock Contention\t119.99\t1.51 Mlocks/sec, 4 threads
\tMemory Test\t86.28\t
\t\tSystem\t72.26\t
\t\t\tAllocate\t575.80\t375.60 Kalloc/sec
\t\t\tFill\t43.17\t343.65 MB/sec
\t\t\tCopy\t60.17\t300.85 MB/sec
\t\tStream\t107.05\t
\t\t\tCopy\t101.45\t741.62 MB/sec [altivec]
\t\t\tScale\t100.39\t740.86 MB/sec [altivec]
\t\t\tAdd\t108.99\t697.56 MB/sec [altivec]
\t\t\tTriad\t119.41\t729.62 MB/sec [altivec]
\tQuartz Graphics Test\t103.29\t
\t\tLine\t95.91\t2.44 Klines/sec [50% alpha]
\t\tRectangle\t111.77\t7.86 Krects/sec [50% alpha]
\t\tCircle\t115.54\t2.66 Kcircles/sec [50% alpha]
\t\tBezier\t95.58\t1.04 Kbeziers/sec [50% alpha]
\t\tText\t100.83\t1.64 Kchars/sec
\tOpenGL Graphics Test\t69.39\t
\t\tSpinning Squares\t69.39\t48.56 frames/sec
\tUser Interface Test\t108.46\t
\t\tElements\t108.46\t34.89 refresh/sec
\tDisk Test\t62.49\t
\t\tSequential\t75.62\t
\t\t\tUncached Write\t77.67\t32.38 MB/sec [4K blocks]
\t\t\tUncached Write\t56.05\t22.95 MB/sec [256K blocks]
\t\t\tUncached Read\t154.90\t24.52 MB/sec [4K blocks]
\t\t\tUncached Read\t63.60\t25.70 MB/sec [256K blocks]
\t\tRandom\t53.24\t
\t\t\tUncached Write\t47.05\t0.71 MB/sec [4K blocks]
\t\t\tUncached Write\t50.51\t11.39 MB/sec [256K blocks]
\t\t\tUncached Read\t56.79\t0.37 MB/sec [4K blocks]
\t\t\tUncached Read\t60.73\t12.50 MB/sec [256K blocks]
Originally posted by Henriok
First.. we know nothing about the 750VX. Second.. the G4 has only about 10% higher performance than the G3 when doing regular FPU work at the same frequency.
Well, from a simplified perspective, the g3 must execute each double precision floating point instruction twice. The g4 does not. So if the code were purely double precision FPU work, then indeed I would presume a big gain on the g4.