So a desktop CPU will have no problem performing a ton of tasks, while a mobile SoC that gets roughly the same score in "Geekbench" will gag, choke and die a miserable death if you hit it with that same ton of tasks. Not to mention that up until recently there were still mobile OSes that couldn't properly perform multitasking. Mobile SoCs are usually a lot worse at this then desktop CPUs. Not to mention that an SoC needs to be updated every time they change wireless standards or network speeds, since the logic for all that telephony stuff is also integrated. And some tasks will make them look "choppy". This is why new versions of the mobile OSes won't work so well on older phones/tablets sometimes. That makes a SoC look very powerful at the time of their launch on the market, but in a very short amount of time (~2-3 years), a lot of those functions will be deprecated, and all that silicon will become dead weight. Mobile SoC also have dedicated logic that perform accelerated tasks, and usually there's a lot of them.
Like for example, encryption or video decoding. Now, some CPUs implement dedicated logic to perform some vastly accelerated tasks, but not many. And this pure performance is almost never achieved in normal use. When running a benchmark, the mobile SoC will ignore it's power consumption and it will configure itself to favor pure performance. So you can't compare "Geekbench" scores with other "Geekbench" scores.Īs the scores you have are probably gathered over different versions of the benchmark, they could be incomparable.Īlso, a desktop CPU is not power constricted as a mobile SoC is. The problem with Geekbench is that it has more then one version and every version performs a different set of tests. If we want to convert that to a high end cpu from an earlier time (which was the original goal), then we are going to need to look for more obscure passmark scores). So the cpu in the iphone 6 is about equivalent to a low grade cpu from 2009. The E7500 gets ~1900 on passmark, so that it close enough for me. Something close to 80% of the passmark performance of the 5350 would be ~2000, so that is what we are aiming for. Meanwhile, a q6600 scores ~3000, so I was right about it being a good bit faster. We want something that is close to 80% as powerful (2900/3500=~83% I know it isn't linear, but we are playing fast and loose here) We can use the 5350 as a reference for further investigation. So we are looking at something about half as powerful. Going off of just the multicore score, the cpu in the zenfone scores ~2900, right around the same as the iphone 6 (it averages higher, but we are just going to call them equal for the sake of convenience). I am not too familiar with mobile stuff, which means that someone more familiar probably would have realized this sooner, but Geekbench. I think that I found something worth investigating. Meanwhile, I don't think that anythign from the nineties would be as fast either, giving a rather large gap in time (and then the vertical movements of the price point) and no way for me to narrow it down. I think that by the time we got to stuff like the Q6600, the iphone would be left in the dust. So yeah, can you guys help me figure this out? My guess is some high end part from the early 2000s. And comparing the flops of gpu and cpu is not fair in any stretch of the word. i can find it for the gpu section, but not for the cpu. The problem is that I can't find the gflops for the cpu portion of the iphone. That would work well enough for rough estimations of processing power in a situation like this. My first thought was that it would be hard to get decent numbers that would be comparable across two platforms over decades. maybe AMD should try to sue them over the name?).Īnyway, she wanted to know what year processor would be equivalent to the cpu in her iphone 6. She has an iphone 6 which (iirc) has an A8 processor (note: not an AMD APU. My girlfriend asked me an interesting question the other day.
0 kommentar(er)
