Alright Sup Forums, I came up with a simple way to calculate a CPU's performance by only getting a few important data, such as: > Number of Cores > Number of Threads > Max Turbo Boost It works with Intel and AMD. The formula is: OPS = { [(C × 2) + T] × MTF } × GC
Where: > OPS = Overall Performance Score > C = Number of Cores > T = Number of Threads > GC = Generation Constant (see pic related) I didn't think it would work so well, it's obviously not 100% precise, but it comes really close.
Let me show an example of it's precision: > i3 8100 (4c/4t, 3.6GHz) - 54 OPS > R3 1300X (4c/4t, 3.7GHz) - 46.6 OPS (50.4 OC) Using the formula the 8100 is 15% faster than the 1300X (7% faster if the R3 is OC'd). Now let's check some benchmarks: Geekbench (Multicore): > i3 8100: ~15000 > r3 1300X: ~14000 (OC) Here, the i3 8100 is 7.14% faster than the 1300X OC'd, which matches with what we just previewed with the formula. Cinebench R15 (Multi-Thread): > i3 8100: ~568 > 1300X: ~515 (non-OC) That gives the i3 a 10.3% advantage, again, really close to the 15% previewed by the formula.
>ryzen lower than haswell gee wonder who made this
Ethan Russell
His formula heavily weighs in favor of cores. So AMD actually looks pretty impressive, if it were normalized for gaming performance AMD would look less favorable.
Aiden Green
where do you see that? the 4770K barely passes the Ryzen 3 1300X.
Connor Reyes
look at OP's image
In reality Ryzen's about on par with 6th/7th gen intel in singlethread and still ahead of 8th gen intel on multithread
Chase Wood
That's the generational constant... Read the formula, it's not saying one is better
Mason Roberts
It's basically accounting for AMDs low IPC and higher core count.
Caleb Martinez
... cont
Now let's see how the R5 1600 ($200) behaves against the i5 8400 ($210) Formula: > R5 1600: 90.7 OPS (100.8 if OC'd to 4GHz) > i5 8400: 90.0 OPS So according to the formula, the 1600 OC is around 11% faster than the i5, now let's check the benchmarks: Geekbench (Multicore) > R5 1600 (OC): ~25500 > i5: ~22500 That's a 11.7% advantage for the 1600
That's a 17% advantage for the R5 1600. It's kinda off what the formula previewed, but still really close.
Pic related is a list I've made of several CPU's, ordered by price.
Christopher Hernandez
A i5 4460 beats the R3 1200, even though they have virtually identical specs.
Ryzen apparently suffers a lot when Hyper Threading is disabled.
Camden Murphy
OP here, GC is not only IPC, it's calibrated to match benchmarks and even gaming. If you give a GC of 6th/7th gen Intel to Ryzen, the results will massively favor Ryzen to the point it doesn't even match AMD's own benchmarks.
Just make the calculations and see for yourself.
Ian Gonzalez
Now, let’s see what the formula previews about the AMD’s golden chip, the R7 1700 ($290), against the similarly priced i7 7700: > R7 1700: 124.3 OPS (134.4 OPS if OC’d to 4GHz) > i7 7700: 82.3 OPS
That gives the R7 a huge advantage of 33.8% stock, and 38.8% OC over the i7. Now, let’s go for the benchmarks.
Geekbench (Multi) > R7 1700: ~31000 (OC) > i7 7700: ~19500 That’s a 38% advantage for the OC’d 1700, just like the formula!
That’s a 38.6% advantage for the R7 1700 (vs the 33.8% previewed by the formula). That’s pretty close considering the formula is an overall performance previewer, it doesn’t need to ace specific benchmarks.
Levi Davis
So, according to the formula, these are the best chips, price/performance wise: > i3 8100 > R5 1600 > R7 1700 > i7 8700K (If OC'd) > TR 1920X > TR 1950X
AMD has better deals on more categories, but Intel has a great fucking low end chip (i3 8100) and I bet the new Pentium will be great as well.
The overclocked i7 8700K performs very nicely, but it's really expensive.
Jack Hernandez
Yeah, that's pretty close. The i3-8100 is pretty much a last gen i5 for an i3 price. That's a pretty damn good deal.
Parker White
You didn't calculate in the power usage.
Charles Smith
>The i3-8100 is pretty much a last gen i5 for an i3 price This.
I hope the new Pentium is quad-core as well, I'm thinking something like 4c/4t at 3.2GHz would be good enough.
Charles Young
I an instruction uses multiple execution units, are those used in parallel, or one after the other?
Ian Jenkins
That's just a Overall Performance Score, but you could easily calculate that like this: PW = OPS/TDP
where PW = Performance/Wattage. The bigger the result, more power efficient the CPU is.
Of course you shouldn't use the advertised TDP, so you would need to make some research before.
Charles Rogers
>new Pentium Wasn't Intel killing off Pentium to boost i3 sales? Or the other way around?
Reminder that if Intel manages to sell the i7 8700 (non-K) for $300-310, it will beat the R7 1700.
I'm obviously not including mobo prices etc.
Anthony Baker
I'm not even at quarter load
Juan Rogers
wat
Ryder Jones
what the fuck is this gay shit
Benjamin Miller
shouldn't you also have to factor average cycles per instruction among other shit?
Matthew Murphy
Overclocking will speed up that portion that's being calculated bud.
Isaac Brooks
that’s already accounted for
Gabriel Morris
An easy way to calculate the performance of a CPU without having to do much research.
Mason Butler
How so?
Jaxon Brooks
GC is that and more
Andrew Diaz
Show how you calculated, or cite your sources for those numbers, because I'm not biting.
There's a real honest, objective formula already out there for calculating raw performance from a processor, factoring operations per second, clock speed, and more. Don't reinvent the wheel.
Adrian Gomez
Yeah, the i5 beats the r3 to crashing your OS.
Connor Lee
How is that calculated? It just lists the different constants. Surely it is not just arbitrary?
Alexander Wright
It's not arbitrary, it's based on clock speed and other actual characteristics of the processor.
David Ramirez
However, I'm not sure if these are even actually computed, because he just kind of brought the numbers out without citing where they're from, or giving the formula used to obtain those numbers.
Jordan Mitchell
Of course they're not actually computed, OP just pulled them out of his ass.
