• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

AMD Ryzen Thread: Affordable Core Act

I'm looking to upgrade my old i5 2500k with a Ryzen 5 1600 and a new mobo and RAM.

My current specs are:

i5 2500k OC to 4GHZ
Biostar T277B Z77 Motherboard
16GB DDR3 1600mhz RAM
AMD Gigabyte RX 480 w/ 8GB GDDR5

The only thing I would keep out of the above system is the Rx480, which is an alright performer and is suited well enough for my needs. But I am thinking about ditching the Motherboard, CPU and RAM with a Ryzen 5 1600, whatever socket of mobo the CPU uses and 16+GB of DDR4. Does anyone think the upgrade would be worth it?

Yeah, that's definitely gonna be a nice upgrade.

I'm thinking of doing the same in the next couple of months (also on an OCed i5 2500k).
 
I'm looking to upgrade my old i5 2500k with a Ryzen 5 1600 and a new mobo and RAM.

My current specs are:

i5 2500k OC to 4GHZ
Biostar T277B Z77 Motherboard
16GB DDR3 1600mhz RAM
AMD Gigabyte RX 480 w/ 8GB GDDR5

The only thing I would keep out of the above system is the Rx480, which is an alright performer and is suited well enough for my needs. But I am thinking about ditching the Motherboard, CPU and RAM with a Ryzen 5 1600, whatever socket of mobo the CPU uses and 16+GB of DDR4. Does anyone think the upgrade would be worth it?
I replaced my 2500K @4.5 with a 1700x@4.0. Definitely an upgrade.
 
I replaced my 2500K @4.5 with a 1700x@4.0. Definitely an upgrade.

I went from an i5 3570k (with a rather mediocre 4.3 Ghz max OC) also to a 1700x@4.0. And yes, a very noticeable improvement, mostly in how much smoother everything feels. My old processor always felt like it was just hanging on in games, and I would get stuttering and bad 99th percentile frametimes.

Can't wait until Vega is officially announced and benchmarked so I can decide which GPU to upgrade to. Hurry up AMD!
 
I'm looking to upgrade my old i5 2500k with a Ryzen 5 1600 and a new mobo and RAM.

My current specs are:

i5 2500k OC to 4GHZ
Biostar T277B Z77 Motherboard
16GB DDR3 1600mhz RAM
AMD Gigabyte RX 480 w/ 8GB GDDR5

The only thing I would keep out of the above system is the Rx480, which is an alright performer and is suited well enough for my needs. But I am thinking about ditching the Motherboard, CPU and RAM with a Ryzen 5 1600, whatever socket of mobo the CPU uses and 16+GB of DDR4. Does anyone think the upgrade would be worth it?
Look at the bottom portion of this post: http://neogaf.com/forum/showpost.php?p=233606587&postcount=2365


When you have some time, have a look at the wide range of R5 reviews listed here:

http://www.neogaf.com/forum/showpost.php?p=233774369&postcount=2444
http://www.neogaf.com/forum/showpost.php?p=233904549&postcount=2530



Don't forget that the X chips have up to a 20C offset, so they're really 20C cooler under full load than is being reported. "85C" is really only 65C.
With an NH-D15, my 1700X at 3.9GHz and ~1.35V hits "73C" when running Intel Burn Test.
This is actually problematic for me because the Crosshair VI does not offset the temperatures for Q-Fan control, and if you exceed the maximum temperature the fans run at full speed.
The highest you can set the maximum is 75C so it's dangerously close to running the fans at full speed under heavy load. Once the ambient temperature gets a few degrees higher, that's probably going to happen.
Correct, I have two monitoring software which report both actual and offset temp values. What I neglected to mention is that there were some odd program results which appeared to be stability-related, and possibly closer to what you'd see from a faulty/failing PCB than a bad CPU OC (also observed at stock 1700X clocks).

The problem was actually a confluence of hardware and software issues which had nothing to do with the motherboard PCB and components or the CPU. Bad program versions + a GPU which is damaged. Mind, the die and VRM are all perfectly fine, but a freak accident apparently damaged the PCB itself. Backup build with a backup GPU so I didn't come across this sooner.

Two of the benchmark programs I use had recent updates which are either bugged across the board or present issues on Ryzen. While running the CPU temps and power draw would obviously be higher which looked to be stability-related. Going back to previous versions solved the issue; no Ryzen-specific problems either.

Antivirus would not complete a full scan, causing system "lockup" similar to Ryzen's Black Screen of Death during bad OCs. Games would not launch, crash altogether, or also have Ryzen "Black Screen of Death"-like results. Also, both the Intel and Killer lan ports had issues similar to bad OCs or failing components.


The GPU did not really display the more common graphical glitches and stability gremlins since the die and VRM are in good order.



·feist·;235182466 said:
Considered starting a thread with this since it's good info for anyone looking to build a system, but wasn't sure it was thread-worthy:

PCWorld —— Intel expects CPU prices to fall now that AMD's Ryzen is here
My guess on this is Intel will just offer better volume deals to OEMs, while us off-the-shelf buyers may not see the benefit.
That's likely to be the most logical result, though the introduction of a consumer 12-core should allow them to at least slightly restructure their line up in an effective price reduction across the board. As opposed to slotting above the current 10-core MSRP, the X299 12-core has to drive the other products lower down somewhat.

If nothing else, there's always contra-revenue and other business practices Intel have employed to gain or maintain market share, though that hasn't always worked out well:

The Motley Fool —— Intel Corp.'s Contra-Revenue Strategy Was a Huge Waste of Money

Unlike tablets, laptops are absolutely crucial for all parties. As far as anyone can tell, Ryzen mobile should be even more competitive than the desktops parts from a performance and efficiency standpoint, before factoring in costs. Intel won't take that lightly.
 

gt86

Member
I just built a new gaming cpu this weekend. I have the Ryzen 1700, gigabyte aorus gaming K5 motherboard, gskill trident z rgb ddr4-3000, aorus rx580. Huge upgrade from my fx8320 and 660gtx. Only issue I am having is my ram will only run at 2100, if I turn on xmp is boot loops. Anyone else having this problem?
 

Paragon

Member
I just built a new gaming cpu this weekend. I have the Ryzen 1700, gigabyte aorus gaming K5 motherboard, gskill trident z rgb ddr4-3000, aorus rx580. Huge upgrade from my fx8320 and 660gtx. Only issue I am having is my ram will only run at 2100, if I turn on xmp is boot loops. Anyone else having this problem?
Make sure that you have the latest BIOS installed.
Don't use XMP. Set the memory speed and voltage manually. (1.35V)
Depending on the RAM you have, 2933MT/s may not be possible right now, so try gradually increasing speed and see what is stable.
 

gt86

Member
Make sure that you have the latest BIOS installed.
Don't use XMP. Set the memory speed and voltage manually. (1.35V)
Depending on the RAM you have, 2933MT/s may not be possible right now, so try gradually increasing speed and see what is stable.

I'm on the newest bios, I've read there are better beta ones out but I can't find them and not so sure I'd want to flash a beta bios. Thanks for the info!
 

masterkajo

Member
Just finished my Ryzen build (apart from the graphic card which will arrive sometime this week).

R5 1600
Asus Prime X370 (Bios 604)
16GB Gskill F4-3200C15D-16GTZ RAM (Samsung b-die)

The RAM worked out of the box @3200 by selecting DOCP profile but the 15-15-15-15-35 timing where changed to 16-15-15-15-35 in windows. I then manually oced the ram by changing the timings to 14-14-14-14-34, setting 1,35V for DRAM voltage and 0,675V (half of DRAM) for the other ram voltage (forgot what it is called). SOC voltage was set to 1.1V and everything worked flawlessly. Apart from the fact that it seems my 1600 is not a good overclocker. I did not manage to achieve a 4GHz or even a 3,9GHz overclock even with 1,4V (more should not be used for a 24/7 OC for years to come). I managed to boot into Windows @ 4GHz with 1,4V but it crashed with prime95 within the first 20 seconds. 3,85 runs fine with 1,35V and 3,8 seems to be stable with 1,3V. I will try to lower the voltage some more for 3,8GHz and see what 3,7 needs. Apparently there is a huge jump in voltage from 3,7 upward.

I am a bit disappointed so far especially since I am planning to pair the 1600 with a 1080ti. And I want to oc the 1600 as far as possible in order to minimize the CPU bottleneck.
 

ESBL

Member
Just finished my Ryzen build (apart from the graphic card which will arrive sometime this week).

R5 1600
Asus Prime X370 (Bios 604)
16GB Gskill F4-3200C15D-16GTZ RAM (Samsung b-die)

The RAM worked out of the box @3200 by selecting DOCP profile but the 15-15-15-15-35 timing where changed to 16-15-15-15-35 in windows. I then manually oced the ram by changing the timings to 14-14-14-14-34, setting 1,35V for DRAM voltage and 0,675V (half of DRAM) for the other ram voltage (forgot what it is called). SOC voltage was set to 1.1V and everything worked flawlessly. Apart from the fact that it seems my 1600 is not a good overclocker. I did not manage to achieve a 4GHz or even a 3,9GHz overclock even with 1,4V (more should not be used for a 24/7 OC for years to come). I managed to boot into Windows @ 4GHz with 1,4V but it crashed with prime95 within the first 20 seconds. 3,85 runs fine with 1,35V and 3,8 seems to be stable with 1,3V. I will try to lower the voltage some more for 3,8GHz and see what 3,7 needs. Apparently there is a huge jump in voltage from 3,7 upward.

I am a bit disappointed so far especially since I am planning to pair the 1600 with a 1080ti. And I want to oc the 1600 as far as possible in order to minimize the CPU bottleneck.

Seems like you got a bad 1600, I've been reading most people are able to get to 3.9 on stock cooling.
 

Datschge

Member
·feist·;235381015 said:
Unlike tablets, laptops are absolutely crucial for all parties. As far as anyone can tell, Ryzen mobile should be even more competitive than the desktops parts from a performance and efficiency standpoint, before factoring in costs. Intel won't take that lightly.
Indeed. Laptops and servers, both markets where 14LPP is perfectly suited. With the vocal focus on gaming AMD kind of picked the most contested least suitable market to launch Ryzen in. While this led to a more subdued hype than Ryzen arguably deserved, all the added scrutiny ensures more efforts to optimize for the new platform by all the technology stakeholders in the market, something that will pay off for everybody involved the more further Zen based products are being released.

As for Intel not taking it lightly, I'm really looking forward to how they are going to handle the continuing disruption. Unlike back during the Pentium 4 debacle Intel doesn't appear to have an alternate CPU design approach they can pull out to reset their fortunes. On the contrary, for nearly a decade Intel was content with rather small but steady improvements where scaling the improvements to more cores for higher end systems has been a time-consuming manual process. Now they'll be more and more forced to do both at once, increase the amount of improvements and make their so far premium many core products more mainstream, all while keeping their insane profit ratio to satisfy their shareholders (something that so far kept killing all their entries in new markets in their infancy due to comparatively too little profit).

Interesting times, to say the least.
 

Seronei

Member
Just finished my Ryzen build (apart from the graphic card which will arrive sometime this week).

R5 1600
Asus Prime X370 (Bios 604)
16GB Gskill F4-3200C15D-16GTZ RAM (Samsung b-die)

The RAM worked out of the box @3200 by selecting DOCP profile but the 15-15-15-15-35 timing where changed to 16-15-15-15-35 in windows. I then manually oced the ram by changing the timings to 14-14-14-14-34, setting 1,35V for DRAM voltage and 0,675V (half of DRAM) for the other ram voltage (forgot what it is called). SOC voltage was set to 1.1V and everything worked flawlessly. Apart from the fact that it seems my 1600 is not a good overclocker. I did not manage to achieve a 4GHz or even a 3,9GHz overclock even with 1,4V (more should not be used for a 24/7 OC for years to come). I managed to boot into Windows @ 4GHz with 1,4V but it crashed with prime95 within the first 20 seconds. 3,85 runs fine with 1,35V and 3,8 seems to be stable with 1,3V. I will try to lower the voltage some more for 3,8GHz and see what 3,7 needs. Apparently there is a huge jump in voltage from 3,7 upward.

I am a bit disappointed so far especially since I am planning to pair the 1600 with a 1080ti. And I want to oc the 1600 as far as possible in order to minimize the CPU bottleneck.

High clocked RAM might hurt your OC, might want to test 2133mhz and see if you can go higher on the cores.
 
Just finished my Ryzen build (apart from the graphic card which will arrive sometime this week).

R5 1600
Asus Prime X370 (Bios 604)
16GB Gskill F4-3200C15D-16GTZ RAM (Samsung b-die)

The RAM worked out of the box @3200 by selecting DOCP profile but the 15-15-15-15-35 timing where changed to 16-15-15-15-35 in windows. I then manually oced the ram by changing the timings to 14-14-14-14-34, setting 1,35V for DRAM voltage and 0,675V (half of DRAM) for the other ram voltage (forgot what it is called). SOC voltage was set to 1.1V and everything worked flawlessly. Apart from the fact that it seems my 1600 is not a good overclocker. I did not manage to achieve a 4GHz or even a 3,9GHz overclock even with 1,4V (more should not be used for a 24/7 OC for years to come). I managed to boot into Windows @ 4GHz with 1,4V but it crashed with prime95 within the first 20 seconds. 3,85 runs fine with 1,35V and 3,8 seems to be stable with 1,3V. I will try to lower the voltage some more for 3,8GHz and see what 3,7 needs. Apparently there is a huge jump in voltage from 3,7 upward.

I am a bit disappointed so far especially since I am planning to pair the 1600 with a 1080ti. And I want to oc the 1600 as far as possible in order to minimize the CPU bottleneck.

Might try a different cooler for the cpu. I don't know what you are using.
 
My brothers got an i7 4790, I tested out rendering some shots in maya on it and it rendered in 93 seconds. Tested out the same shot on my newly built ryzen 7 computer (first time building a computer) and it rendered it out in 44 seconds. That's over twice as fast. Damn, REALLY glad I got the ryzen 7. Its basically two i7's for the price of one.

Only problem is that the 16 GB ram I have is only going at 2133 even though I got 3000, but its not an issue for what I'm doing since from what I hear faster RAM only makes a significant difference when you have a lot of geometry, but I work with low poly stuff. That's the BIOS problem right? I hope its fixed in the future. I'll try getting the latest bios and see if that makes a difference.
 

Kataploom

Gold Member
My brothers got an i7 4790, I tested out rendering some shots in maya on it and it rendered in 93 seconds. Tested out the same shot on my newly built ryzen 7 computer (first time building a computer) and it rendered it out in 44 seconds. That's over twice as fast. Damn, REALLY glad I got the ryzen 7. Its basically two i7's for the price of one.

Only problem is that the 16 GB ram I have is only going at 2133 even though I got 3000, but its not an issue for what I'm doing since from what I hear faster RAM only makes a significant difference when you have a lot of geometry, but I work with low poly stuff. That's the BIOS problem right? I hope its fixed in the future. I'll try getting the latest bios and see if that makes a difference.

What Ryzen 7 model did you get? Just to have an idea, I've been hesitating between getting an i7 instead of Ryzen due to the maturity of the platform but it's mostly for work on 3D, video and stuff, tbh
 

masterkajo

Member
Seems like you got a bad 1600, I've been reading most people are able to get to 3.9 on stock cooling.

Might try a different cooler for the cpu. I don't know what you are using.
I will try with my Hyper212+ again today. Was testing with a stock cooler and temps were going up high 75°+ but I put it on 100% fan speed to compensate. Really quiet at that speed to be honest.

High clocked RAM might hurt your OC, might want to test 2133mhz and see if you can go higher on the cores.
Good point. Will put the RAM back and see how far the processor can go on its own. Should have thought of this myself. But it was very late yesterday.

OK let's see what I can achieve today. 3,9GHz should hopefully be doable at least. Just out of curiosity, how much difference does a 100 to 150MHz difference do regarding FPS with a 1080ti? Because most people say it is not worth to push for 4GHz since most of the time the GPU will be the bottleneck @1440p anyway (if I buy a 1070 for example) but nobody seems to be as stupid as I am and pairs the R5 1600 with a 1080ti.
 
I went from an i5 3570k (with a rather mediocre 4.3 Ghz max OC) also to a 1700x@4.0. And yes, a very noticeable improvement, mostly in how much smoother everything feels. My old processor always felt like it was just hanging on in games, and I would get stuttering and bad 99th percentile frametimes.

Can't wait until Vega is officially announced and benchmarked so I can decide which GPU to upgrade to. Hurry up AMD!
Yeah, definitely less stuttering for me as well.

I was planning to buy my wife a Vega and move her 480 to my PC and Crossfire, but she's pregnant now so that's not happening. Gotta start spending less and prepare our budget for kiddo.
 

llien

Member
I keep hearing the "less stuttering" (actually even at games and when doing nothing in parallel and even vs 7700k in a game like GTA5), which makes me wonder, is it something measurable/visible in tests?
 

Renekton

Member
I keep hearing the "less stuttering" (actually even at games and when doing nothing in parallel and even vs 7700k in a game like GTA5), which makes me wonder, is it something measurable/visible in tests?
Yeah some sites like Techreport give frametime graph.
 

Mr Swine

Banned
Well I changed my order from a 1700x to a 1600x so I can put that money on a new camera. 6 cores and 12 threads on the 1600x is not that bad even if the 1700x will probably be faster in the future.
 
I keep hearing the "less stuttering" (actually even at games and when doing nothing in parallel and even vs 7700k in a game like GTA5), which makes me wonder, is it something measurable/visible in tests?

You'd notice it via frametimes if the FPS of a game was demonstrating stuttering. As for GTA 5 in particular, the reason why people say that game stutters more than most on Intel is because of a bug in the game engine, it craps itself above 165 FPS. It was quite noticeable on Intel, particularly the latest i5's, but not as noticible on older CPU's or on Ryzen because those CPU's weren't generally fast enough to push the engine to the breaking point, though you can replicate the engine bug on Ryzen. Hyperthreading seems to mitigate the issue on the i7's.

http://www.gamersnexus.net/guides/2878-gta-v-stuttering-mystery-on-i5-r5-cpus-part-2 - Gamersnexus goes into detail into GTA 5 stuttering.
 
Agner Fog's Ryzen testing and optimisation update:

http://www.agner.org/optimize/
http://www.agner.org/optimize/blog/read.php?i=838
http://www.agner.org/optimize/microarchitecture.pdf



·feist·;235094713 said:
RAM & CPU overclocking "workshop" direct from AMD's Robert Hallock:

*** Blunty [YouTube] —— We Got Inside AMD’s Texas Home - & Learned How To Overclock RAM on Ryzen ***

Extended 52 minute-long presentation video:

RedGamingTech [YouTube] —— MSI & AMD Dragon Squad | Robert Hallock’s Entire Talk on Ryzen & Overclocking

Very much recommend owners watch either of those videos above.




https://twitter.com/iBUYPOWER/status/857741376770850818

Interested in Ryzen? Here's a few usage models 😉 Learn more about our Ryzen PCs! http://www.iBUYPOWER.com/Ryzen

C-dOtKOUMAE5Ky1.jpg:orig



Indeed. Laptops and servers, both markets where 14LPP is perfectly suited. With the vocal focus on gaming AMD kind of picked the most contested least suitable market to launch Ryzen in. While this led to a more subdued hype than Ryzen arguably deserved, all the added scrutiny ensures more efforts to optimize for the new platform by all the technology stakeholders in the market, something that will pay off for everybody involved the more further Zen based products are being released.

As for Intel not taking it lightly, I'm really looking forward to how they are going to handle the continuing disruption. Unlike back during the Pentium 4 debacle Intel doesn't appear to have an alternate CPU design approach they can pull out to reset their fortunes. On the contrary, for nearly a decade Intel was content with rather small but steady improvements where scaling the improvements to more cores for higher end systems has been a time-consuming manual process. Now they'll be more and more forced to do both at once, increase the amount of improvements and make their so far premium many core products more mainstream, all while keeping their insane profit ratio to satisfy their shareholders (something that so far kept killing all their entries in new markets in their infancy due to comparatively too little profit).

Interesting times, to say the least.
Agreed. Apart from any possible actions on Intel's part you also have to account for inertia across the board from vendors and OEM's reluctance to adapt, to consumers' heavy reliance on brand recognition above the actual viability of the product.

Mid-level to high-end Ryzen mobile will be severely hampered if the parts are primarily available in the same single channel entry level low quality chassis with poor 1366x768 or 1600x900 displays AMD is often consigned to.
 
·feist·, thanks for posting the Blunty video about RAM overclocking. There were definitely a few very interesting tidbits in there regarding what voltages AMD considers safe and tweaks that have worked for them in testing.
 

Datschge

Member
·feist·;235557133 said:
Agner Fog's Ryzen testing and optimisation update:

http://www.agner.org/optimize/microarchitecture.pdf
Excellent! (For those wanting to dive right into the parts specific to Ryzen, branch prediction is on page 34, pipeline on page 211, and kernel on page 223.)

·feist·;235557133 said:
That's a nice way to depict the performance differences.

·feist·;235557133 said:
Agreed. Apart from any possible actions on Intel's part you also have to account for inertia across the board from vendors and OEM's reluctance to adapt, to consumers' heavy reliance on brand recognition above the actual viability of the product.

Mid-level to high-end Ryzen mobile will be severely hampered if the parts are primarily available in the same single channel entry level low quality chassis with poor 1366x768 or 1600x900 displays AMD is often consigned to.
Inertia will always be a problem indeed. But Ryzen is poised to make a significant difference in all relevant parameters: price, performance and efficiency. That should enable AMD and other stakeholders in the market to make Zen based products sufficiently stand out among competing products, and market competitors not following suit with Zen based products of their own will have a hard time matching them without neglecting one or more of the parameters. Availability may actually be the biggest issue then.

And AMD's constant slow drip feed of new Zen based chips works in its favor as well so far, allowing each set to stand on its own for some time which further increases market awareness of its qualities. (That's also an interesting contrast to the botched publicity handling of AMD's graphics unit that I guess is more concerned with non-public facing development like for APUs and semi-customs rather than high-end dedicated graphics.)
 

Steel

Banned
Since PUBG has been big lately I thought this was interesting:

A comparison between the 1800x and I7 6800k in PUBG. Unfortunately it uses all ultra settings at 1440P so it could be GPU-bottlenecked(well it's a 1080), but the results are interesting nonetheless, both get 61 FPS average, but the minimums are different, 33 FPS min for the I7, 46 FPS min for the 1800x. Surprisingly PUBG seems to use all the threads on Ryzen at least a little.

In my experience, if I turn down GPU intensive settings(ultra AA, everything else low-med) but keep draw distance at max on my 1700 at 3.9 ghz paired with my GTX 970 I get anywhere from 70-130 FPS at 1080P(Well 40-50 in the beginning lobby where everyone is, clusterfuck that it is), so this scenario is definitely GPU bottlenecked. Putting all ultra puts me in the 45-60 range, 25-40 in the lobby. If someone could find a scenario that isn't gpu bottlenecked, I'd be interested. I really wanna know how much my 970 is holding me back at this point.

That being said, turning down shadows in battlegrounds seems to actually help with spotting people, oddly enough.
 
Since PUBG has been big lately I thought this was interesting:


A comparison between the 1800x and I7 6800k in PUBG. Unfortunately it uses all ultra settings at 1440P so it could be GPU-bottlenecked(well it's a 1080), but the results are interesting nonetheless, both get 61 FPS average, but the minimums are different, 33 FPS min for the I7, 46 FPS min for the 1800x. Surprisingly PUBG seems to use all the threads on Ryzen at least a little.

In my experience, if I turn down GPU intensive settings(ultra AA, everything else low-med) but keep draw distance at max on my 1700 at 3.9 ghz paired with my GTX 970 I get anywhere from 70-130 FPS at 1080P(Well 40-50 in the beginning lobby where everyone is, clusterfuck that it is), so this scenario is definitely GPU bottlenecked. Putting all ultra puts me in the 45-60 range, 25-40 in the lobby. If someone could find a scenario that isn't gpu bottlenecked, I'd be interested. I really wanna know how much my 970 is holding me back at this point.

That being said, turning down shadows in battlegrounds seems to actually help with spotting people, oddly enough.


Interesting indeed.

It's curious that GPU usage is not really maxed out, which suggests there might be some CPU bottlenecking at ultra settings (increased view distance increasing draw calls maybe?).

 
Glad you guys found the OC and Agner links useful.


That's a nice way to depict the performance differences.
There's a Korean tech site which included that style of diagram in their Ryzen 7 1800x, 1700x and 1700 review, along with standard bar graphs. I meant to post it at the time, though never got around to it. The site has since reviewed the Ryzen 5 lineup.

I'll post them later.


Inertia will always be a problem indeed. But Ryzen is poised to make a significant difference in all relevant parameters: price, performance and efficiency. That should enable AMD and other stakeholders in the market to make Zen based products sufficiently stand out among competing products, and market competitors not following suit with Zen based products of their own will have a hard time matching them without neglecting one or more of the parameters. Availability may actually be the biggest issue then.

And AMD's constant slow drip feed of new Zen based chips works in its favor as well so far, allowing each set to stand on its own for some time which further increases market awareness of its qualities. (That's also an interesting contrast to the botched publicity handling of AMD's graphics unit that I guess is more concerned with non-public facing development like for APUs and semi-customs rather than high-end dedicated graphics.)
I'd like to see what if any relationship there may be between Ryzen mobile and the Ryzen 3 lineup, or if R3's later launch is in part due to needing to amass a large inventory of chips from binned 8-core dies.

IIRC, Ryzen desktop will not go below 4c/4t which leaves room for 2c/4t mobile parts and lower-priced Athlons, if that brand will continue being used. All in all, with competent leadership and competitive processors AMD are in a good position to address the monopoly that exists from the consumer space to enterprise.

Even Intel-only users and companies will benefit from performance at lower prices, or higher performance at same prices.




Bangkok Post: Tech —— AMD to power up chip market


Advanced Micro Devices (AMD) has resumed marketing chips for gaming computers after a five-year absence in Thailand.

The US-based chipmaker seeks to cash in on a segment that is still seeing high growth, especially the high-end gaming computer market, which has defied the overall continuous drop of the PC market.
AMD has also widened its server processor products for data centres and cloud computing platform, which are also high-growth areas.

"We aim to regain our market share in chips for the server processor field, where the company has enjoyed up to a 20% market share in the past," said Mr Jirawat.

Sasiprapa Sutheerapat, business development manager of AMD's Thailand branch office, said at least five major international notebook makers are scheduled to roll out their notebooks equipped with AMD Ryzen notebook processors in the second half of this year.
 

IC5

Member
Just finished my Ryzen build (apart from the graphic card which will arrive sometime this week).

R5 1600
Asus Prime X370 (Bios 604)
16GB Gskill F4-3200C15D-16GTZ RAM (Samsung b-die)

The RAM worked out of the box @3200 by selecting DOCP profile but the 15-15-15-15-35 timing where changed to 16-15-15-15-35 in windows. I then manually oced the ram by changing the timings to 14-14-14-14-34, setting 1,35V for DRAM voltage and 0,675V (half of DRAM) for the other ram voltage (forgot what it is called). SOC voltage was set to 1.1V and everything worked flawlessly. Apart from the fact that it seems my 1600 is not a good overclocker. I did not manage to achieve a 4GHz or even a 3,9GHz overclock even with 1,4V (more should not be used for a 24/7 OC for years to come). I managed to boot into Windows @ 4GHz with 1,4V but it crashed with prime95 within the first 20 seconds. 3,85 runs fine with 1,35V and 3,8 seems to be stable with 1,3V. I will try to lower the voltage some more for 3,8GHz and see what 3,7 needs. Apparently there is a huge jump in voltage from 3,7 upward.

I am a bit disappointed so far especially since I am planning to pair the 1600 with a 1080ti. And I want to oc the 1600 as far as possible in order to minimize the CPU bottleneck.
Yeah, try the RAM stock. That will ease the memory controller and may allow a better CPU overclock.
Then try to push the RAM again. See if you can get 3000. That's sort of a performance sweet spot right now. Even if you have to relax the timings a bit, 900 extra mhz will give a good boost.

However, you may find that maxing your ram and leaving you Ryzen stock, is overall faster than squeezing your Ryzen for 100 or 200 more mhz, if you cannot also get your RAM to 3000.

My brothers got an i7 4790, I tested out rendering some shots in maya on it and it rendered in 93 seconds. Tested out the same shot on my newly built ryzen 7 computer (first time building a computer) and it rendered it out in 44 seconds. That's over twice as fast. Damn, REALLY glad I got the ryzen 7. Its basically two i7's for the price of one.

Only problem is that the 16 GB ram I have is only going at 2133 even though I got 3000, but its not an issue for what I'm doing since from what I hear faster RAM only makes a significant difference when you have a lot of geometry, but I work with low poly stuff. That's the BIOS problem right? I hope its fixed in the future. I'll try getting the latest bios and see if that makes a difference.
Dis you try to set XMP in your bios?
DDR3000 is over standard AMD spec. Motherboards usually do not default to your RAM's highest advertised speed. You will either have to set XMP, or try to manually over lock the ram, to that level.
If XMP does not work, its possible that RAM is not a good match for your motherboard and you may not have good manual overclocking results. Check the official RAM list for your motherboard.
 
May Microcode Update for Improved DDR4 Support - AGESA 1006 (1005)/1.0.0.6 (1.0.0.5)



Some items specific to Gigabyte, others apply across the board to every motherboard maker's impending updates:


http://forum.gigabyte.us/post/4419/thread

GIGABYTE - Matt:

Hi guys,

Wish I could drop in and give you guys a new BIOS but I don't have them yet
EeK2FQVh3Ae7YoVg0RdQ.png
Latest word is they are working on a new set with AGESA 1006... I am pushing as hard as I can to get you guys something new to test as soon as its ready. Just wanted to say that we are working on it (I know that phrase gets old - trust me I am a user/gamer/enthusiast too) but me going to Dreamhack (for work not as a participant) doesn't stop the BIOS team or engineers from working on the code ;-)

Just to recap these are the issue being worked on:

*For those looking for IOMMU fixes we are hopefully going to have an option to force boot off a specific PCIe slot. Its not the grouping fix, but a work around for now.
*Disable LAN (Per request)
Disable Audio (Per request)
"ROM Image update" (Being worked on with AMI, no ETA)
Cold boot / Wont boot. Have to re-flash BIOS. (people have referred to this as "soft brick")
*AGESA 1006 -  improve memory (Got high hopes for this one. Going to enable 20+ memory register)

(Edit) P-State overclocking / downclocking while overclocked.

* Should be included in next round of BETA BIOS - but no promises, just the latest I heard from our BIOS guys.




http://forum.gigabyte.us/post/4434/thread

GIGABYTE - Matt:

Thank you for the update. And could you tell the team, that some more options/settings related to memory (procODT settings, bootvoltage of ram etc) or in general OC would be awesome to get?


I'll pass it along, but right now #1 priority is memory.

You bring up an interesting point about the new code... Here's a little background info. When AGESA 1000 was around we did a ton of work to certify memory. Then we got AGESA 1003, then 1004A (which was a patch on top of 1003). 1004a was a 2 steps forward 1 step back process. Many of the 'fixes' we implemented were broken by 1004a. On top of that we received the binary code - not the source code - which means we couldn't see exactly what was changed. This is why the previous set of BETA BIOS had "memory compatibility" as an option. It allowed users to switch back and forth from AGESA code to see which worked for their particular memory. 

Much of the work in the previous BETA BIOS was fixing things that were previously working... My understanding is that AGESA 1006 should unify a lot of these issues.




http://forum.gigabyte.us/post/4446/thread

GIGABYTE - Matt:

thanks for the headsup. Are they also aware about the power management issues discussed a few pages back?


Ah, ya, P-states on the list as well.




http://forum.gigabyte.us/post/4531/thread

GIGABYTE - Matt:

Looks like AMD decided to go with AGESA 1005 after all... Current ETA for BETA BIOS is next week. I'll post it the moment I get it.




http://forum.gigabyte.us/post/4548/thread

GIGABYTE - Matt:

what does that mean for RAM compatibility? wasn't 1006 gonna add all those RAM kits?
Do you know what the changes are between 1004a and 1005?


1005 opens up 20 memory registrars, which should help with memory compatibility.
 
·feist·;235938877 said:
·feist·;235557133 said:
That's a nice way to depict the performance differences.
There's a Korean tech site which included that style of diagram in their Ryzen 7 1800x, 1700x and 1700 review, along with standard bar graphs. I meant to post it at the time, though never got around to it. The site has since reviewed the Ryzen 5 lineup.

I'll post them later.
Here:

Graphs embedded for quick reference, along with vast compilation of Ryzen 7 and Ryzen 5 reviews to cross-reference (featuring both Nvidia and AMD GPUs to highlight current performance discrepancies with Nvidia's driver not having been optimised for Ryzen*, compared to Intel optimisations, whereas AMD drivers have optimisations for Intel & Ryzen).

*see AnandTech review game benchmarks beginning here with 2 Nvidia -vs- 2 AMD GPUs* (will update once testing shows Nvidia has addressed Ryzen optimisation)




Ryzen 7 Reviews
http://www.neogaf.com/forum/showpost.php?p=231347427&postcount=602
http://www.neogaf.com/forum/showpost.php?p=231397257&postcount=788
http://www.neogaf.com/forum/showpost.php?p=233736067&postcount=2417 (Audio Production)

ComputerBase compiled: http://www.neogaf.com/forum/showpost.php?p=234291529&postcount=2644

TechSpot - Best CPUs: http://www.neogaf.com/forum/showpost.php?p=234373297&postcount=2650

The Tech Report - System Guide (March '17): http://www.neogaf.com/forum/showpost.php?p=232666379&postcount=2028




DrMOLA —— CPU WARS: EPISODE VIII - THE LAST REBEL (Ryzen 7 1800x, 1700x, 1700 Review) [Korean]


Left: R7 1800x -vs- i7 6900k -vs- FX-9590
Center: R7 1700x -vs- i7 6850k -vs- i7 6800k
Right: R7 1700 -vs- i7 7700k -vs- FX-8370e

Review contains below diagrams and performance bar graphs

2kkg4.jpg
b8kvs.jpg
fgkyg.jpg


***Graph translations via 3D Center: here, here, here***

***Use translations as key for untranslated R5 graphs below***





Ryzen 5 Reviews
http://www.neogaf.com/forum/showpost.php?p=233774369&postcount=2444
http://www.neogaf.com/forum/showpost.php?p=233904549&postcount=2530

ComputerBase compiled: http://www.neogaf.com/forum/showpost.php?p=234291529&postcount=2644

3DCenter compiled: http://www.neogaf.com/forum/showpost.php?p=233887127&postcount=2526

TechSpot - Best CPUs: http://www.neogaf.com/forum/showpost.php?p=234373297&postcount=2650




DrMOLA —— [CPU] Meet Zen At a Much Cheaper Price - AMD Releases Ryzen 5 (Ryzen 1600x, 1600, 1500x, 1400 Review) [Korean]

DrMOLA —— [CPU] The Ebb-Effect Begins: The Economics of Ryzen 5, Ryzenomics (Ryzen 1600x, 1600, 1500x, 1400 Review) [Korean]


Left top: R5 1600x -vs- i7 6800k -vs- i7 7700k
Center top: R5 1600 -vs- i7 7700 -vs- i5 7600k
Right top: R5 1500x -vs- i5 7600 -vs- i5 7500

Review contains below diagrams and performance graphs

523bc28e54b636bc54c656xip.jpeg
4b274324c1231ab6423agczkx.jpeg
0bda17f5d901af064d2fn2ytr.jpeg


Left mid: R5 1600x -vs- R7 1700
Center mid: R5 1600x -vs- i7 6850k -vs- R5 1600x OC
Right mid: R5 1600 -vs- i7 6800k -vs- R5 1600 OC



Left bot: R5 1500x -vs- i7 7700 -vs- i5 7500 -vs- R5 1500x OC
Right bot: R5 1400 -vs- i5 7600 -vs- i5 7400 -vs- R5 1400 OC

 
With all the debate regarding whether games will utilize more cores/threads in the near future, I was curious, where does that support actually get implemented, the engine itself? For example, does Epic have to build that into the next version of Unreal 4, and then any game using that version of UE4 automatically take advantage of the added cores without the developer having to be mindful of that sort of optimization? Or is it at a much higher level of the programming and is something that will vary from game to game or even version to version?
 

Datschge

Member
·feist·;236178576 said:
Thank you ·feist·, very interesting comparisons. I guess the 100% hexagon is based on R7 1800X except for the last set of light blue graphs where it's based on R5 1600X?
 
With all the debate regarding whether games will utilize more cores/threads in the near future, I was curious, where does that support actually get implemented, the engine itself? For example, does Epic have to build that into the next version of Unreal 4, and then any game using that version of UE4 automatically take advantage of the added cores without the developer having to be mindful of that sort of optimization? Or is it at a much higher level of the programming and is something that will vary from game to game or even version to version?
I know that in UE3 & 4, there are specific engine config files where you can set the amount of cores/threads to be utilized. With UE4, IIRC, the engine is set up to use all but one core by default, while in UE3, you have to go into the BaseEngine.ini or (insert game name)Engine.ini file to set the amount of cores shaders can use.
 
for what exactly do you need a 650W PSU on the 1600build. im running virtually the same build with 450W overclocked without any issue at all....
 
New Tech Report system guide out, was pleased to see a lot of AMD representation in it. I've used their guides as a base in the past (with a few tweaks) with good results.
 
I've added a Ryzen build to the PC build sheet. I've tried to be concise but also include mentions of what to do about BIOS and memory.
I'm unsure of what higher feature board to add apart from the two smaller mATX for best value.

Let me know what you think.
http://www.neogaf.com/forum/showthread.php?t=1352338

05/10/2017: Ryzen build added! Full information on Ryzen and Motherboard BIOS and RAM compatibility [Guru3D Thread]

I noticed that you are pretending that HEDT doesn't exist. If you're going to add a build for 6/8 core Ryzen, you are doing a disservice to anyone who actually uses your guides for building by failing to at least mention the existence of Intel's 6/8 core desktop platform.
 
I noticed that you are pretending that HEDT doesn't exist. If you're going to add a build for 6/8 core Ryzen, you are doing a disservice to anyone who actually uses your guides for building by failing to at least mention the existence of Intel's 6/8 core desktop platform.

For the sake of Gaming--which I assume is the focus of his guides--the HEDT platform offers very little. An i7-7700k is the best performance you're going to get, so you'd basically be advising people to pay a lot more money for lower performance in games. They are impressive chips for things like Encoding but really terrible Price/Performance for Gaming.

EDIT: On further inspection it probably should be listed under Enthusiast, and maybe under "More Cores!"
 

Hazaro

relies on auto-aim
for what exactly do you need a 650W PSU on the 1600build. im running virtually the same build with 450W overclocked without any issue at all....
Additional efficiency and or multi GPU setup. It's not default.
New Tech Report system guide out, was pleased to see a lot of AMD representation in it. I've used their guides as a base in the past (with a few tweaks) with good results.
Was hoping to cheat with item recs, but nope. Oh well.
I noticed that you are pretending that HEDT doesn't exist. If you're going to add a build for 6/8 core Ryzen, you are doing a disservice to anyone who actually uses your guides for building by failing to at least mention the existence of Intel's 6/8 core desktop platform.
They can post in the thread if they are doing a more specialized build and dropping 2.5k+. Builds are building blocks for discussions, but strong recommendations.
You're more than welcome to put one together and I'll add it into the OP and give your post a hyperlink.
 

spyshagg

Should not be allowed to breed
I noticed that you are pretending that HEDT doesn't exist. If you're going to add a build for 6/8 core Ryzen, you are doing a disservice to anyone who actually uses your guides for building by failing to at least mention the existence of Intel's 6/8 core desktop platform.

The disservice would be to recommend an Intel HEDT. They exist in the market, but you should not buy one at the current pricing, ever.
 
Top level Ryzens leaked:
http://wccftech.com/amd-ryzen-16-12-core-x399-whitehaven-cpus-leaked/

So....I cant wait to read how i7700k is still 1.2fps better on GTAV if you downscale the resolution to 480p with a GTX1080ti. :D
Judging by how similar the 8 core Ryzens and the 6 core Ryzens have been in games, articles going "16 core Ryzens are shit at games for the money" is an inevitability.

Although, frankly, I'd expect people to use these 16 core chips for home servers and virtual machines, not gaming.
 

Durante

Member
For the sake of Gaming--which I assume is the focus of his guides--the HEDT platform offers very little. An i7-7700k is the best performance you're going to get, so you'd basically be advising people to pay a lot more money for lower performance in games.
I don't think this is true in many modern games any more.

The 6 core HEDT chips are pretty interesting in terms of gaming (if you don't mind the price). You can get them up to ~4.4 GHz which gives excellent sequential performance, and you also have 6 cores / 12 hardware threads which is pretty much at the scaling limit for most games.
 

Mr Swine

Banned
So how accurate is Ryzen Master? When I play GTAV it reports temperature around 45 Celsius with my Noctua NH-U12S cooler. It's around the same with Starcraft 2.
 
So how accurate is Ryzen Master? When I play GTAV it reports temperature around 45 Celsius with my Noctua NH-U12S cooler. It's around the same with Starcraft 2.
Should be most accurate.
There was an issue where all temperature reading apps were reporting 20° higher than the chips were running. Ryzen Master was updated to show actual temp.
 
Top Bottom