Why don't AMD fix their single core performance?

Why doesn't your mom fix her single porn performance?

File: 1485557444734.jpg (311KB, 640x1136px) Image search: [Google]

311KB, 640x1136px

>>60022946

People dont get one thing.
We got single core performance so high that hypertreading was introduced to make a fucking use of all that power and rather spread that performance between more threads.

File: 1315690645265.jpg (269KB, 1006x850px) Image search: [Google]

269KB, 1006x850px

>>60022946
single core is bretty gud when you compare clock-to-clock performance from jewtels offerings.

http://www.agner.org/optimize/blog/read.php?i=49#829

File: 1492893633027.png (158KB, 2033x909px) Image search: [Google]

158KB, 2033x909px

Why hasn't Intel improved its single thread performance for 4 years?
Do you seriously think Intel will be able to hide behind higher clockspeeds on low core parts for more than a year?

>>60023297
Just wait for graphene, Intel will once again take the lead with 10GHz

>>60023461
I'm waiting for their unobtanium based CPUs. Gonna hit 100 THz on air.

>>60023224
Wow, this is brutal.

If Agner fucking Fog tells you something like this, Intel is truly damn fucked.

>>60023505
FTL Quantum cores will be able to compute into the future and obtain answers even before they were queried

>>60023224
>http://www.agner.org/optimize/blog/read.php?i=49#829
>The single thread IPC is higher than that of anything intel
is this guy retarded or just ignorant?

>>60025862

Neither: http://www.dtu.dk/english/service/phonebook/person?id=78696&tab=2&qt=dtupublicationquery

Insanely smart and highly trusted in the industry for his words due to long history of reliable contribution to the computing industry.

>>60022946
because from bulldozer until very recently, AMD was banking on every application becoming extremely multi-threaded *and* calling on the CPU to handle that parallel workload. Something which hasn't....really happened, sure there's a use for more than 4c/8t in workloads like video production, but we're also seeing a lot of the workloads that would require such parallelization being offloaded to the GPU more and more, since those are already vastly more parallel. A trend that'll probably continue, leading to 8c/16t being the sweet spot for a while to come.

>>60026477
I'm still expecting 6c/12t to be the new standard.

>>60025862
>He doesn't know about Agner "Jew Wrangler" Fog

http://www.agner.org/optimize/blog/read.php?i=49#49

>>60026504
Of course it will, intel has nowhere else to go until next architecture and amd is still very focused on multithreaded performances. And when old games like GTA V are nearly bottlenecked by i5s, it's pretty obvious that the consumer industry will have to shift to more cores. It's also more power efficient than intel 4.7GHz housefires so you can expect laptops to up the core count too.

>>60026372
>>60026525
Except he is completely wrong

>>60027299
Says you.

>>60027299
Except he's not

File: 2017-03-26 11_38_16.png (15KB, 403x402px) Image search: [Google]

15KB, 403x402px

>>60022946
why doesn't intel fix their multi core performance?

>pic related: $330 CPU vs $1600 CPU

File: 1489160516428.png (245KB, 800x612px) Image search: [Google]

245KB, 800x612px

>>60022946

>>60027827
>>60027873
Ryzen does not have more IPC than anything intel has to offer

>>60023024
From what I understand, hyperthreading only kicks into effect once more threads are being used than you have physical cores. If you're playing a game designed for a quad core CPU on a quad core CPU without hyper threading, the CPU will use all four cores without hyper threading instead of using two cores with hyper threading.

>>60028725
Shit, I meant quad core CPU WITH hyperthreading.

>>60022946
>AMD just killed entire INTEL division
NVIDIA is next

>>60028764
>AMD just killed entire INTEL division
Shit I can really feel their market share still being above 60% what is intel gonna do

>>60022946
Because only poorly made programs rely on single threaded performance nowadays. The future is packing as many cores in as possible while maintaining an acceptable level of single threaded performance for legacy applications.

>>60028775
you realize that's 30% less than last year?
for giants like intel it's a disaster, the bigger you are the bigger every little blow becomes

File: Bait_This_Is_Sakura.png (261KB, 694x761px) Image search: [Google]

261KB, 694x761px

>>60028705

>>60023297
honestly, I love the fact that cpu performance has stagnated because now I don't feel the need to upgrade for a very long time

>>60028824
Single thread will never disappear. Not every program can be parallelized, even in theory, and writing parallel code is much more expensive. Even parallel engines rely on a master loop to keep things sync'd. The future is only in parallel CPUs because we've hit a wall in single core performance. Taking advantage of 2-4 cores isn't hard but 8? Yeah... very tough thus DX12 to get something out of it.