AMD Zen 2 (Ryzen 3000) - *** NO COMPETITOR HINTING ***

[Hotwired]

You can have a good look on my Ghetto ddr cooling solution hehehe

But why

 

[Troezar]

There's 2 sites selling it, but yes it's actually a pre-order i guess. The price is quite high compared to cheaper 3200 kits but not as bad as i thought it would be. I don't know how reliable either site is but it looks like one might be a bit better than the other.. (doh)

Be interesting to see how it compares with the Ballistix Sport LT kits.

 

[Zeed]

But why

cause i tend to pump 1.5v in to ddr4 on 24/7 working system?? Do You know how temperature effects memory erors on ryzen ??

 

[Hotwired]

cause i tend to pump 1.5v in to ddr4 on 24/7 working system?? Do You know how temperature effects memory erors on ryzen ??

Fair enough, I didn't think there was much value in pushing memory hard enough to need active cooling. It's really rare to see it.

 

[Zeed]

Fair enough, I didn't think there was much value in pushing memory hard enough to need active cooling. It's really rare to see it.

Thats why i was first on OCN to get 3466cl14 stable on 1700x pass 45c on modules you get errors I'm still waiting for my laster thermometer to check how hot they run with fan nowadays
Yuri added to newer versions of Ryzen calculator warning that extreme and high speed pressets REQUIRE active cooling on ddr.

 

[Zeed]

Anyhow if anyone wants to have a GOOD read about memory fine tuning on Ryzen Yuri wrote this article
https://www.techpowerup.com/review/amd-ryzen-memory-tweaking-overclocking-guide/

Good watch this was

 

[Scougar]

Hit 40C in my garden in Essex, UK. That’s hot enough for this country!

40c? You sure your thermometer wasn't in the sun, as temps are measured in the shade.

Was 37C today and humid hell. Running around in a paintball mask. Think I got some footage lol.

I think I am gonna wait a while.to.see how the ryzen stuff pans out. Seems to need a little time to mature and bring some more mainstream boards party.

 

[Perfect_Chaos]

How are people finding their GPU covering some of the chipset fan? looks like it happens with a few boards even the Asus TUF x570. Dumb positioning, but if it makes little difference then i guess it won't matter.

 

[Perfect_Chaos]

Or is the Gigabyte aorus elite better for the money?

 

[michty_me]

Or is the Gigabyte aorus elite better for the money?

I had heard the VRM's were on par with the Taichi.

 

[chrcoluk]

I find it just plain odd that they felt the need to change it at all, how come they didn't change it when AMD pushed out the Ryzen lineup in 2017, or when Intel started doing 18c/36t HEDT CPU's to counter Threadripper, why now?

2% - do you really think that in a line up of modern many core processors that 2% of the total weighting is correct?

Look at Intel's current line-up of CPU's in the 9th Gen listing's only 13 are 4-cores, and 30+ are 6-cores or more, that is over 230% more CPU's with higher core counts. It makes no sense for the 'user' who is using the data, completely ignoring the manufacturer.

The timing is odd I agree on that, right after 3000 series launch. But as I been saying in many posts, I have never ever used the cpu ranking on their site before, and I have also never used the effective speed value, that value has always been based on their opinion. I just use the site to compare raw test results between cpu's, and ignore their weighting system. For that purpose absolutely nothing has changed on the site.

For everyone who is interested I have almost finished my testing of 2600X vs 3700X vs 8600K. The conclusion is in about 2-3 months I am probably going to buy a 3700X, so you can maybe predict the results from my decision.

What I have been testing includes.

Microsoft office
VLC player
Chrome, IE, Firefox browsers.
Modern action games such as division 2 and GTA5.
Legacy games such as emulators, starcraft 1, modded FF7, and WC3.
Modern ported JRPG's such as the tales games, FF13 series including the ultimate test lightning returns, and FF15.
Driving games such as project cars, and autosport
Strategy games such as civilisation series, warhammer series, age of empire games, C&C series and more.
Bench tools such as cinebench, aida, passmark, 3dmark, pcmark and XTU
Windows responsiveness tests.
Linux compiling tests.
Bare metal virtualisation tests. Ryzen loves this workload.
GIMP

The workload that I havent really bothered with is content creation (except GIMP as I use it for creating mods). I will post a report of my findings sometime next month. Ryzen 3000 did well for itself, Intel is still faster on a 4.7ghz 8600k (I downclocked my 8600k as I feel 4.7 is the highest realistic clock to use) in some of these tasks but the advantage is no longer what I feel a decision maker, and the areas where the Ryzen 3000 comes out on top in some cases is very dominant. In addition the Ryzen 3000 chip does its performance on average 34% less power consumption and 18C lower temps. Note also those power/heat savings are after the downclock (and also large down voltage I removed 40mv, so would be even worse for the 8600k @ 4.8).

 

[Zeed]

Or is the Gigabyte aorus elite better for the money?

Thats what I got for my mate's build quite liked it.

 

[TheLegendOfMart]

I know that the Ryzen 3 3200G isn't Zen2 but Zen+.

I've been playing around in Ryzen Master and got it up to 4.1GHz from 3.6Ghz and APU from 1200 to 1625 and everything seems stable at just under 75c using OCCT stress test using stock voltage and a cheap air cooler.

This is my first Ryzen and I've never overclocked one before, do these values seem safe to run at?

 

[Minstadave]

I had heard the VRM's were on par with the Taichi.

They are, the Elite looks great value.

 

[L0rdMike]

I know that the Ryzen 3 3200G isn't Zen2 but Zen+.

I've been playing around in Ryzen Master and got it up to 4.1GHz from 3.6Ghz and APU from 1200 to 1625 and everything seems stable at just under 75c using OCCT stress test using stock voltage and a cheap air cooler.

This is my first Ryzen and I've never overclocked one before, do these values seem safe to run at?

Depends on what sort of voltage you are using.

 

[TheLegendOfMart]

Stock voltage, it hovers around 1.45v.

 

[michty_me]

They are, the Elite looks great value.

Even better then!! I'm seriously looking at the Elite or the Pro over the Taichi now.

 

[ShortWarning]

Installed my 3700x into my msi carbon x370 over the weekend. No issues with boot or anything etc.

I ran a quick real bench and monitored with Ryzen master, it would only boost to 4.1. This seems quite a way off the advertised clocks. Temps got to 72 degrees under a kraken x62 (Ryzen Master says that 90 is the limit). Is there anything I can do?

 

[Zeed]

Download: in the next half hour the archives will be added on all resources
Techpowerup link
Guru3d link
Сomputerbase.de

Changelog 1.6.0:

• Added full support for Zen 2 and X570 motherboards.
• Added recommendations for VDDG and FCLK for Zen 2.
• Added recommendations for Vref (CHA / CHB).
• The maximum frequency for 1:1 mode is now 3800 MHz, and for 2:1 mode - 4200.
• Updated presets for Samsung b-die, Hynix CJR and Micron e-die.
• Added backward compatibility of Zen 2 with previous generations of motherboards.
• Added PMU Training recommendations for all Ryzen platforms (AM4/TR4). Better training - a more stable system.
• Added autosave of the main tab settings when closing the program and automatically loading them when the program is started.
• Improved tWRRD prediction for Dual Rank.
• Color notification of voltage limits and FCLK. Black is the norm, red is the limit.
• Timing detection for Zen and Zen + generations. For the generation Zen 2 at the moment there is no possibility to add a definition of timings due to NDA. Unfortunately.
• Fixed a lack of presets for X399.
• Fixed a rare error in which it was impossible to calculate the timings.
• Added a button to access the Internet overclocking statistics generation Zen 2.
• For users of Intel systems in the “Additional Calculators” section, it is possible to calculate tREFI and RTL IO-L.
• Added 2 new latency tests for MEMbench. Custom latency is a memory access test that the processor cannot predict (using a specific offset). Random latency - the average time is measured for random memory accesses in a buffer of a certain size. Partially affects the result of TLB. Both versions are release candidates, but I do not deny that there will be calibrations.
• Custom latency and Random latency are included in the Easy and Default packages. Please note that during testing the program will open / close the prompt itself (you do not need to deal with it).
• Added information about the size of the L3 cache processor. Also, the size of the test blocks for latency tests is controlled relative to it.
• Added button Clear standby. There is to clean the cached memory by the operating system. Didn't the test run? - Click Clear standby.
• Max RAM button better determines the amount of free memory available for testing. Useful for checking memory for errors.
• There are exceptions for many functions. Fault tolerance of the program increased.
• Graphic edits.

I want to thank @slafniy and @A_z_z_y for helping to create this version.

BR, Yuri (@1usmus)

 

[opethdisciple]

AMD are really missing a trick here by not having their own tool to calculate memory timings.

Still very impressive work by 1usmus on his dram calculator.