@ Starscream When I restart my computer later Ill underclock it and run the tests again.
@NaturalViolence I have an n330 with ion crammed into an NES case. Ill get around to running the test on it eventually.
(02-11-2013, 06:46 AM)NaturalViolence Wrote: [ -> ]I can imagine a lot of people giving me funny looks as I run around best buy running GC/Wii games on all of their PCs.
There's two tricks to get around this.
1. Only test on one PC per visit.
2. If testing on two or more, say you're comparing the performance of the computers.
Also, at least for Walmart, they have USB mass storage disabled. HOWEVER! You can still read from an SD card just fine.
That's a lot of effort to put into this for very little gain.
Up until 2 years ago I only had an Athlon XP, so I very much WAS testing the performance.
In case anyone is curious so far the calculated IPC of the different architectures for this benchmark are (normalized to trinity and rounded to two decimal places):
Ivy bridge: 1.55-1.65
Sandy bridge: 1.40-1.50
Phenom II: 1.12-1.12
Athlon X2 Brisbane: 0.70-0.74
Oh and I guess we have SS's llano results. I'll add those shortly once I finish the calculations.
If anyone wants to do it themselves the formula is F(x,y,A,B) = (Y/B)/(X/A)
where X is gamespeed of first cpu, Y is gamespeed of second cpu, A is clock rate of first cpu, B is clock rate of second cpu. And yes I can establish a proof for that. I really don't want to though. I would use GS sub 0 and GS sub 1 for gamespeed and CR sub 0 and CR sub 1 for clock rates but I have no idea how to type that notation on here.
TBH IPC isn't really a very fair measurement when something like Trinity runs at higher clocks than Llano but consumes less power (on the laptop side anyway) even though both are 32nm.
All I care about is performance. Since each family of cpus has many different models with different clock rates it's important to know the relationship between clock rate and performance relative to other architectures.
Anyways the llano results are 1.08-1.13. Just barely lower than phenom II, which I expected. I'm actually kind of surprised that my predictions based on review site benchmarks were so accurate. Looks like I don't have to change my hierarchy thread so far.
Either way, your IPC numbers seem EXTREMELY similar to each CPU's single-threaded scores in Cinebench R11.5 - heck even Trinity gets a score around 1.00 in Cinebench R11.5.
Was this intentional?
No but it makes sense considering cinebench is often used to measure single threaded performance as well as IPC.
But IPC can differ largely based on what is used by the program. In particular, Cinebench is supposedly very floating-point heavy.
Nevertheless, this could theoretically mean that we could use the single-threaded results in Cinebench R11.5 as a "cheat sheet" for Dolphin performance.