Rene Lamontagne wrote:
On 04/22/2019 8:31 AM, Yrrah wrote:
Rene Lamontagne :

Intel i7 8700 CPU $400

Why Intel, why not an AMD Ryzen CPU?


Yrrah


I haven't used an AMD CPU for many years but when I did use them they
were A,OK and I am sure they are even better now, the chip sets were the
reason I went to Intel, I suppose they too are much improved now.
I know The Intel may cost more but I am more familiar with them and they
have not given me any problems.
As far as video cards My choice is AMD over Nvidia
BTW what would be a Ryzen equivalent to an Intel i8700 Coffee lake?

Thanks, Rene

You can work this out using cpubenchmark.

AMDs premise, is to give you more on multithreading,
rather than have the highest clock*IPC on single threaded
benches. Games need some of each. "Highest clock" for
the boss thread, "Moar cores" to the extent that other
parts of the game (AI, map prefetch) can be parallelized.

https://www.cpubenchmark.net/singleThread.html # single threaded, highest clock
# The most common operations use this.

Intel Core i7-8700 @ 3.20GHz 2,630 # $420 CDN
AMD Ryzen Threadripper 2920X 2,231 # $800 CDN, mobo++, large socket
AMD Ryzen 7 2700X 2,193 # $420 CDN

https://www.cpubenchmark.net/high_end_cpus.html # "towing capacity"
# 7zip goes faster, movie enc goes faster

Intel Core i7-8700 @ 3.20GHz 15,153 # $420 CDN
AMD Ryzen Threadripper 2920X 22,044 # $800 CDN, mobo++, large socket
AMD Ryzen 7 2700X 16,985 # $420 CDN

AMD clock*IPC cannot quite match Intel clock*IPC.
That's why the single thread results are lower.

AMD sometimes has a higher clock at the same price, but
the IPC (Instructions Per Clock) is a bit lower, and so
head-to-head single thread, AMD loses.

AMD compensates by trying to use more cores.

Usually on AMD, when the core count is extremely high,
the bench results become confused. For example, I might
expect to see a higher 7ZIP bench, with all that silicon
sitting there, and the results are instead, disappointing.
Then I don't know if this is Microsofts fault (scheduler),
Igors fault (wrong compiler), or AMDs fault
(flimsy connection to DRAM channels).

Anyway, those two pages are a start, but not the entire story.
There are always particular benchmark tests, where one of
the players falls flat. "Single point measurements" like
the above, are flawed, but they're easy to write up
in a USENET post.

Paul