AMD Ryzen 7 1700 Review: The New King of Price To Performance Ratio

[Valentyn]

42 minutes ago, N3v3r3nding_N3wb said:

Honestly, I wish Vegas would've turned out better and been more popular.  I don't do any video editing, but Adobe being the only giant in that segment hurts people I do care about and like watching.

 

Adobe are lazy bastards with their performance optimization. So much so a little MacBook Pro with Final Cut Pro X will beat a custom Windows workstation most of the times. 

Why? Apple actually use as many cores as possible, the GPU, RAM , and can be set to actively render in the back via GPU/IGPU as you're currently editing and working.
HardwareCunucks did an interesting video comparing it all, and he even disabled the Background Rendering for his tests, which would have given FCPX  an even better advantage over Premiere Pro.
 

Imagine if Adobe even got a 50% increase in performance for the amazing hardware we have available.

 

[Deli]

18 hours ago, Daniel644 said:

I didn't know how to OC my 6700k, then I watched a video or 2 and then I knew how, NOT knowing how is a STUPID excuse to pay 170 dollars more

This applies to many things. Like cable sleeving. It costs like $70 - $90 for a set of custom sleeved cable. But it only costs me $30 in material if I do it myself. 

[Doobeedoo]

Well not everyone is willing or know to OC so pay more for higher frequency. Also it can have better OC potential even if it's few 100-200Mhz or so.

It depends from person to person.

[Daniel644]

Just now, Doobeedoo said:

Well not everyone is willing or know to OC so pay more for higher frequency. Also it can have better OC potential even if it's few 100-200Mhz or so.

It depends from person to person.

a 2-5% increase in performance is NOT worth a nearly 50% increase in price

[SpaceGhostC2C]

19 hours ago, Rangaman42 said:

Good to see people comparing the r7 chips to the correct Intel counterparts. People comparing a big 8 core workstation CPU to a smaller gaming focused SKU is getting really old. I know the r7's are priced around 7700k levels, but they're not a direct competitor.  The smaller 4/6c Ryzens will likely overclock better and compete more directly, especially with fast RAM.

The sad part is when they pick these two CPUs... and then  run gaming benchmarks 

17 hours ago, cj09beira said:

here is a nice graph i made from silicon lottery's data

 

https://docs.google.com/spreadsheets/d/1JmupyJ7NaVUevjYxUY-KNllKn6HwP4cG2XCuowrnTzg/pubchart?oid=2002362242&format=interactive

Nice! I miss the confidence bounds, though  Wonder if they provide enough information to compute them...

[Energycore]

19 hours ago, stealth80 said:

Gonna throw this in the pot as well

 

 

Wow that's some gains holy shit

[Taf the Ghost]

13 minutes ago, Energycore said:

Wow that's some gains holy shit

The big Windows 10 patch didn't actually have anything in the background to effect Ryzen.  It did, however, tend to reset any custom settings with Power Management.  There's an issue with Core Parking, which might actually be more of the issue than the Scheduler.  (There's still issues with the Scheduler, but it's really only highlighted when a few other issues crop up.)

 

What we've been steadily seeing is the fact it's a new CPU architecture and chipset.  Little improvements here and there, along with better understanding, and that performance gap between the Workstation & Gaming benchmarks has closed up considerably.  At this point, most of the Day 1 Gaming Benchmarks are not that useful.

 

The other wrinkle that'll work itself out over time is that some Game Engines are just really well-tuned to Intel's quirks, which some of those quirks seems like it causes some Max FPS caps with Ryzen. (I think some games more aggressively try to switch cores, which is the reason why they're benefiting so much from the higher RAM speed, since that drops the latency in the Infinite Fabric.)

[4960X]

so will microsoft release an update to fix ryzen's scheduler or is there nothing wrong with the current scheduler?

[DrMikeNZ]

9 hours ago, 4960X said:

so will microsoft release an update to fix ryzen's scheduler or is there nothing wrong with the current scheduler?

AMD announced a while back that they found no evidence of any issues with the scheduler.

Quote

We have investigated reports alleging incorrect thread scheduling on the AMD Ryzen™ processor. Based on our findings, AMD believes that the Windows® 10 thread scheduler is operating properly for “Zen,” and we do not presently believe there is an issue with the scheduler adversely utilizing the logical and physical configurations of the architecture.
 
As an extension of this investigation, we have also reviewed topology logs generated by the Sysinternals Coreinfo utility. We have determined that an outdated version of the application was responsible for originating the incorrect topology data that has been widely reported in the media. 

https://community.amd.com/community/gaming/blog/2017/03/13/amd-ryzen-community-update

[Valentyn]

10 hours ago, 4960X said:

so will microsoft release an update to fix ryzen's scheduler or is there nothing wrong with the current scheduler?

 

1 hour ago, DrMikeNZ said:

AMD announced a while back that they found no evidence of any issues with the scheduler.

https://community.amd.com/community/gaming/blog/2017/03/13/amd-ryzen-community-update

 

That was in response to SMT issues. PCper found that there are issues with keeping work on the same CCX with the scheduler though. They found that while investigating the SMT scheduler issues.

 

They postulated that an update of some sorts might be required to address it.

 

 

There were also tests ran with one CCX vs two, where they found a near 20% performance difference in Battlefield 1. Recent tests from Hardware unboxed only found a 1-2FPS difference. So it's clear something addressed that issue somewhere.

 

 

[Taf the Ghost]

There will probably be a Windows Scheduler tweak in Windows 11, since it'd be something MS would do for the next revision of Windows Server first.  It's really only a penalty in a few situations, unless you can get your RAM speed super high.  Zen+ or Zen2 should have a much higher clocked Infinite Fabric, which brings the latency down.  The Scheduler issue, that does exist, seems to be more something about spamming hard across the CCX bridge for little purposes, but that seems mostly limited to a few Game Engines.

 

The SMT issue was far more that a bunch of games were reading Ryzen wrong.  That's why there were SMT Off boosts in some games, but wasn't a Scheduler issue.  The big upgrades has been in getting the RAM timings stable.  The Infinite Fabric seems really sensitive to that, along with the rest of the Memory System.  A brand new memory sub-system really takes a lot of work.

 

Though some REALLY interesting stuff is starting to drop.

 

 

Bios revision sees a 5 FPS (5-7% improvement) in Total War: Warhammer, one of the games Ryzen did pretty badly in during the first run of testing.

 

http://www.bitsandchips.it/9-hardware/8200-asus-crosshair-vi-hero-dal-bios-0902-al-1002?start=1

 

It's in Italian, but it's kind of fascinating to watch this play out.  Synthetic benchmarks haven't seemed as sensitive to the issues as games, likely because of the nature of the workload more than anything else.  But, in total, we're starting to get a picture of why some of the early benchmarks were all over the place, beyond just the reviewers not yet knowing how to get the most out of the chips.

[Citadelen]

1 hour ago, Taf the Ghost said:

-snip-

Warhammer is the kind o game that needs a patch, seeing as AMD is the hardware partner for Warhammer, a fix is probably in the works. (I also recall Creative Assembly themselves saying they were working on a patch)

[4960X]

dat feeling when you can't find an x370 mobo in stock

[Taf the Ghost]

 

Testing on the 1800X vs 7700k again, but it's nice to see someone actually dive deeper into the game engine issues for once.  Plus the hilarious bonus of Crossfire RX480s toasting an OC'd 1070.

[Guest Darth Revan]

6 hours ago, Taf the Ghost said:

 

Testing on the 1800X vs 7700k again, but it's nice to see someone actually dive deeper into the game engine issues for once.  Plus the hilarious bonus of Crossfire RX480s toasting an OC'd 1070.

Would people please stop posting video from this idiot.

 

[MandelFrac]

On 29/03/2017 at 3:43 AM, Daniel644 said:

a 2-5% increase in performance is NOT worth a nearly 50% increase in price

That depends who you ask. This is exactly what Amazon does every year upgrading their EC2 offerings and paying thousands of bucks for the same core counts but the 3% increase in performance that came along for Broadwell.

[4960X]

19 hours ago, Darth Revan said:

Would people please stop posting video from this idiot.

 

I don't know who he is but why do you think he is an idiot?

[Valentyn]

8 minutes ago, 4960X said:

I don't know who he is but why do you think he is an idiot?

They disagree with his findings, because they don't like it. Despite two other places having found similar results as in that video.