What's new
  • Please do not post any links until you have 3 posts as they will automatically be rejected to prevent SPAM. Many words are also blocked due to being used in SPAM Messages. Thanks!

Ryzen 5000 Gets an Infinity Fabric BOOST! Rocket Lake Clocks up to 5.5Ghz!

Sagath

Moderator
Staff member
Folding Team
Joined
Feb 7, 2009
Messages
4,910
Location
Edmonton, AB
My hope is 4000 on rev-e as well. Clocking high is it’s specialty. 1:1 made it pointless before though

edit: I assumed you meant micron rev-e and not some Samsung bin. Samsung e is a bit of a wildcard I think.
No, correct, Micron E.

I havnt tried to push it beyond tightening the timings at 3600. Since CL16 is 'default' XMP 3600, I assume I can hit 4000.

[email protected] is 8.9ns, where [email protected] is 9ns, so its slightly 'slower' overall.

1603162492407.png
 

Entz

Well-known member
Joined
Jul 17, 2011
Messages
1,659
Location
Kelowna
Awesome chart! I think it is going to take a bunch of benchmarking for sure. Crucial/Micron runs their own 4000 at 18-19-19-39 so that may be a "safe" target . You will take a latency hit but the speed may offset it.
 

Jokester

Well-known member
Joined
Nov 20, 2007
Messages
472
Location
North Shore
Yeah haven't tried tweaking my 3600 c15 8x4 setup. I know it can clock well but probably a pita trying to get each setting dialed in.
 

Sagath

Moderator
Staff member
Folding Team
Joined
Feb 7, 2009
Messages
4,910
Location
Edmonton, AB
@Sagath question...is the.1ns actually going to be noticed when you have the benefit of 400mhz extra bandwidth between 3600mhz @ 8.9nz vs 4000mhz @ 9ns?
Not how it works.

Slower ram at a faster CL can equal the performance of faster ram at a slower CL. Hence the chart I linked.

Essentially if the chart timings in NS are the same (or very close), functionally the ram is operating the same regardless of 'specs'.

 

lowfat

Moderator
Staff member
Joined
Feb 12, 2007
Messages
10,466
Location
Grande Prairie, AB
The issue I have with that is that it doesn't take any consideration of all other timings. Only clock and cL.
 

Mr. Friendly

Well-known member
Joined
Nov 21, 2007
Messages
5,513
Location
British Columbia
Not how it works.

Slower ram at a faster CL can equal the performance of faster ram at a slower CL. Hence the chart I linked.

Essentially if the chart timings in NS are the same (or very close), functionally the ram is operating the same regardless of 'specs'.

ahhh! well...I can't read that chart because of the colors. my vision is overwhelmed and I cannot focus on any of the text. :p
 

Bond007

Well-known member
Joined
Jun 24, 2009
Messages
5,942
Location
Nova Scotia
Not how it works.

Slower ram at a faster CL can equal the performance of faster ram at a slower CL. Hence the chart I linked.

Essentially if the chart timings in NS are the same (or very close), functionally the ram is operating the same regardless of 'specs'.

Agree. I just want to amplify that the caviat with ryzen is that the fabric clock gets tied to the RAM freq. So while the RAM latency may be the same, the fabric clock being faster will give better performance in some uses. So 3200mhz CAS 16 is the same latency as 3600 CAS 18, the performace on ryzen should be a bit better with the 3600 due to the fabric clocking higher.

Obviously other timings will come into play as well.
 

Sagath

Moderator
Staff member
Folding Team
Joined
Feb 7, 2009
Messages
4,910
Location
Edmonton, AB
Agree. I just want to amplify that the caviat with ryzen is that the fabric clock gets tied to the RAM freq. So while the RAM latency may be the same, the fabric clock being faster will give better performance in some uses. So 3200mhz CAS 16 is the same latency as 3600 CAS 18, the performace on ryzen should be a bit better with the 3600 due to the fabric clocking higher.

Obviously other timings will come into play as well.
@lowfat & @Bond007 Correct. My comment was directly regarding basic RAM configuration, all else being equal.

Adding in secondary timings & Zen creates a totally different scenario of performance capability, and definitely affects overall performance.
 
Top