AMD's Ryzen CPUs (Ryzen/TR/Epyc) & Vega/Polaris/Navi GPUs

3700x looks like a best value buy right now. Both 3900X and 3700X appear to have little to no OC headroom left at all, and some reviewers reporting they're not seeing the advertised turbo clockspeeds or able to get anywhere near them.

 

Yeah, seems around 4.3 all core. But when voltage scaling fails, temp does allow scaling.

Even with that and having lower performance in some games, seems to be good value. But, at estimated all core, even with IPC gains, the 9900K does win when overclocked, depending on task (excluding games).

Bearded hardware will be streaming at 1pm Eastern time today to oc the 3700x and possibly putting it on ln2 live if getting bored with water or a aio.

But, really should look good for CPU benchmarking. Don't have any clue on 3Dmark benching yet.

Great value overall.

 

Hardware Unboxed fried their 3900x. Lived a very short life.

 

As I had mentioned, Intel is not dead. At least not in gaming. Overclocking on the Intel still exists where forget any thing extra from the Zen cores. Now if AMD initial hype of 5.0 GHz plus were actually true it might be a different story, but it is not.

Don't get me wrong, it seems they are the productivity kings just not a game changer, pun intended.

 

The Beard OC's the new Ryzen 3700x Live, it could last for *hours*.

"Slacker Live" Stream 003-Overclocking the new AMD Ryzen 3700x
bearded hardware
539 watching now - Started streaming 59 minutes ago
The normal "Slacker Live" stream will be every Thursday 4pm EST(NYC) . Today we will be benching the new AMD Ryzen 3700x live on the MSI X570 Godlike motherboard.

 

sigh.. in the end as i suspected garbage of an overclock, when AMD doesnt talk about something even when major events gone by, usually means something bad. better than zen+ by about 100-150mhz. hope to see AMD take the same approach as intel. lessen density for Zen3 to allow higher frequency, which they totally can with TSMC's 7nm+ due to higher density than 7nm.

upping that by another 300mhz would finally give us 4.5 to 4.6ghz on all 12 - 16 cores.

honestly, for calculation task, browsing, everything that CAN fit inside L3 without having to go much to ram, AMD decimate intel without a doubt by easily 10%+, the IPC of zen2 dominates. sadly it is held back by it's scale-able design having higher memory latency.

@ajc9988 @tilleroftheearth @Papusan

time to wait for zen3/4 vs ice/tiger lake

 

Cool watching now! If anyone is going to get anything out of these chips it's probably going to be him.

 

in consumer space intel's approach for monolithic die will still win in terms of 90% of performance due to shorter distance thus having less latency with it's on die IMC.

kinda sucks that icelake still on core so the base design is like almost 10 yrs old and it is inferior to zen2. maybe we get to see new and improved tigerlake on new arch that will still be monolithic. I wish for a 12 cores on intel's cpu, or maybe zen3/4 can move MC back onto the die making them IMC again.

 

So, 7/7/2019 is done and mostly anti-climatic. With not much being changed in the overall order for most consumer workloads today. Especially real-world consumer workloads.

Nothing new here, as could have been predicted by anyone with a single rationale bone in their body.

• AMD is showing synthetic performance increases over 'ancient' Intel platforms today. Yawn.
• A 7nm process that is more power-efficient than 14++++++. Yawn.
• Still a 'best-bang-for-the-buck' mentality when pure performance in all aspects is still a pipe dream. Sad yawn.
• Intel is still the performance champ with AMD not delivering a solid knockout blow to be the better all-around platform, in all aspects, period.

On their own, IPC, two or three process node advantages and even PCIe 4.0 and other inane metrics mean squat when you need to get your work done. Elaborate marketing and 'rah-rah AMD' get together's isn't enough either to make a dent in actual workloads that matter to most.

Real-world testing as shown on many sites is still Intel's game when pure 'max # of cores' performance isn't required (and yeah; it isn't by 99.9% of users still, 8 cores/16 threads are still enough for most of today's common workloads, as witnessed in the BM's). Not to mention the glitches not encountered when running an Intel platform vs. (and especially a new) AMD.

Kudo's to AMD for delivering on price and most of what they promised (at least in the synthetic 'scores'). I for one appreciate the competition they bring to the table.

It only took AMD 15 years or so to get back to being in this kind of position (actually competing with Intel), while only being serious for the last half dozen years or so.

Let's see what 2025 will look like (on the AMD side). I believe that is the timeframe that is needed for AMD to get their whole act together (hardware, drivers, and real-world performance that equals what Intel already capably demonstrated it can do, for years now). If AMD can keep this kind of pressure on Intel for the long-term, they truly will be a force to be reckoned with.

But for now, Intel, the ball is in your court again.

 

The AMD 3rd Gen Ryzen Deep Dive Review: 3700X and 3900X Raising The Bar
https://www.anandtech.com/show/14605/the-and-ryzen-3700x-3900x-review-raising-the-bar

The AMD Radeon RX 5700 XT & RX 5700 Review: Navi Renews Competition in the Midrange Market
https://www.anandtech.com/show/14618/the-amd-radeon-rx-5700-xt-rx-5700-review/

AMD seems to be well positioned now in both the CPU and GPU areas.
I was admittedly worried a bit about Navi, but it looks like its trading blows (or comes very close to) 2070 Super... at a lower price no less and better efficiency.
I also suspect we might see improved gaming (and compute) performance with better drivers.

Navi's compute capabilities seem lower, but I kinda expected this (especially since its a mid-range GPU).
I did mention before that if Navi was a modified GCN uArch (or retains GCN at least in some fashion), then one of the ways they could improve it for gaming is to either remove or disable compute performance (by keeping it at a Polaris level) and increase gaming relevant hw, which might also positively reflect on power draw (and this actually doesn't disappoint, because according to total system power draw, Navi is MORE efficient than both 2060/2070 and their super variants - at least people shouldn't be able to complain about that).

Those teraflops numbers don't exactly make too much sense in relation to how the GPU's behave in actual compute... but the article does mention this to be a driver problem:
"The big issue at the moment is that while AMD’s drivers are in fairly good shape for gaming, the same cannot be said for compute. Most of our compute benchmarks either failed to have their OpenCL kernels compile, triggered a Windows Timeout Detection and Recovery (TDR), or would just crash. As a result, only three of our regular benchmarks were executable here, with Folding@Home, parts of CompuBench, and Blender all getting whammied."

Huh... its actually funny how in the past AMD performed admirably in compute, but first day release gaming drivers tended to need a bit of work (which was fixed promptly).

Will this affect RVII prices?
It might, and given time, its price will probably drop anyway, but in a larger scheme of things, RVII is in fact a compute powerhouse (in particular when compared to the entire Turing line - and it does so for a fairly cheap price to boot vs Nvidia)... so, depending on how Navi ends up behaving with better drivers in the compute department, AMD might want to shuffle RVII into prosumer space entirely... make it a cheap (or similarly priced) prosumer card... and realistically, it IS a modified data center GPU.

Hopefully, these will get more people interested in Navi (and seriously, it will be interesting to see what AMD does with it when they use chiplets for GPU's - though by then, its possible their new uArch - the one beyond 'Navi' - will actually be released).

 

It's over now, OC max was 4.4ghz with 3600mhz memory on the 3700x was the limit today - CB15 = 2317, along with TimmyJoe who also got 4.4ghz, 4.5ghz would boot into Windows but CB15 reboots.

The last part was OC'ing the memory to 4000mhz, to be continued tomorrow 4pm-5pm (PT?).

Ryzen 3 seems to respond to OC'ing with an edge up in scores against PB2, but that might be due to needing better cooling as well.

Synthetics are great for the Ryzen 3 CPU's, but the real wins have been with the production comparisons with the 3700x beating out the 9900k @ 5ghz, with of course the 3900x stretching out that lead - minutes saved with both on production runs. The Ryzen 3's have a big uplift in rendering, compression, decompression, Premiere in particular. And, with a few exceptions are within a few percentage points of the 9900k in gaming - not enough difference to notice in actual use, with far less power draw than the 9900k.

Lots of good reviews with good tests and results, you just gotta take the time to watch / read the reviews, or at least skim through them to find the graphs.

OC3D found something interesting, they were able to run PCIE 4.0 SSD (Aorus) at full 4900MB/sec on x470 when paired with a Ryzen 3 CPU - same performance as with the same CPU on the x570. He didn't show graphs or demonstrate, he just passed on the tip, so YMMV. Check out his 3700x / 3900x review toward the end for the tip.

 

So, at [email protected] with mem at 2133, he tied Prema's CB15 run at 5.1GHz on the 9900K. Beat it with mem at 3600@cl18. So there is some there there, just not as much as hoped.

Edit: the closest cb15 scores for 9900K to his 4.4ghz with 3600CL16 mem score of 2317 is Der8urner (not Der8auer) with 5.2GHz at 2315, then a 2321 score using 5.3GHz by RKINSLO on HWBot. Wait for tuning.

 

Might be because they reduced the IF sensitivity to RAM speed and latency on Zen 2... so the results might be showing this in conjunction to using X570 mobo?
We might not see more gains until Zen 2 is tested with much higher RAM speeds, or AMD releases new chipset and modified BIOS support to fully take advantage of the higher speed RAM, but who knows? By then, we might end up with Zen 3 on the table which could further improve on Zen 2 weaknesses (depending on which parts of Zen 2 are considered a weakness - intra-core latency could do some work).

Plus, on the clocks, yeah, it does appear as if the initial projections of 'beyond 5GhZ' seemed to have been a bit overzealous. But this could be down to a few factors which at the time were not immediately detectable:
1. AMD using 7nm for low power and mobile parts (as TSMC 7nm node was divided into high perf. parts node, and mobile parts node - though we were told that AMD will be using a high perf. process this time around, but who knows - they were supposed to use the IBM's high perf. process node for 12nm Zen+ but instead ended up using Samsung's 12nm which is an exact derivative of their 14nm node which GLOFO and AMD used for Zen 1 and was designed for low clocks and mobile parts). Not sure though which one they've ended up using in the end. Last known info in the community was high perf. node... but could be the other one.

2. 7nm (initial production) just isn't suitable for high clocks. Intel had years to refine 14nm after all and bring the clocks up at lower voltages, so, I suppose this is the best we can hope for at this stage. Though 7nm+ (EUV) could improve on this (and hopefully this time without raising TDP further).

 

The Ryzen 3600 and 3600x are looking great in gaming and other tasks, these CPU's are real bargains. There aren't many reviews for these CPU's yet, with some reviewers saying these 3600 / 3600x or 3800x CPU's were not sampled for review...:

#RIPINTEL? AMD Ryzen 5 3600 & 3600X Gaming Benchmarks
Gear Seekers
Published on Jul 7, 2019
It's time to talk about Ryzen 3000 and the performance. We decided that we wanted to cover 2 of the lower specced Ryzen 3000 chips. The reason why is because with the last generation and the generation before the 1600, 1600X, 2600 and 2600x were most peoples entry point. Even for me, my personal favorite chip has been the 2600. So we decided to kick off our Ryzen 3000 coverage with the Ryzen 5 3600 and the 3600X.


Ryzen 5 3600X vs i7-8700K — 30 Benchmarks — A New King?
Tech Deals
Published on Jul 7, 2019
Zen 2 launches today, do we have a new desktop CPU King? Today we compare the 2 best 6 core desktop CPUs across 30 different benchmarks.

 

Yeah, these will do nicely for gamers, and those who want to do some productivity too on the cheap.
Looks like productivity is on par with 2700/2700x while surpassing it in games (now I kinda want Zen 2 3700x in my Helios 500).
AMD could easily capture a good portion of the market with these whilst ensuring at least 1 more upgrade cycle to Zen 3 (possibly beyond if they decide to extend AM4 socket support past 2020 - but I suspect they may want to run AM4 up until Zen 3 and then go with a new socket in 2021 for DDR5 - but they might also decide to release X670 for Zen 3 with existing AM4 and DDR5).

 

Quick run for the hell of it.

 

Sensational launch from AMD. Anandtech are showing an IPC lead over Intel for the first time in 15 years. Zen 2 is the real deal and Intel's entire line-up looks ridiculously bad for the price.

I was pretty surprised with the progress they made with Navi as well. AT showing the 5700 XT 5% behind the 2070 Super. They'd been treading water on the GPU side for way too long. It's not a Turing killer, and Nvidia are still on 12nm so I expect next year to be big for Nvidia, but it is an excellent step in the right direction.

 

AMD's 5700/5700XT built-in HEVC encoding is able to sustain 6 simultaneous streams and this channel has previously been doing NVENC tweaking / OBS tuning, but have happily found the additional - hardly mentioned - streaming facility of the new Navi's:

BUDGET STREAMING ON A WHOLE NEW LEVEL | 3700X + B450 + RX 5700 = Production & RENDERING BEAST
EposVox
Published on Jul 7, 2019
AMD Zen 2 architecture is here! And thanks to Wendell at Level1Techs, I got a chance to get my hands on the Ryzen 7 3700X and Ryzen 9 3900X and put them to the test for content creation, video production, and live streaming - and even look at Navi! We've taken a look at Ryzen 3000 for video editing and general use and the RX 5700 and RX5700XT for streaming and editing, lastly we're taking a look at how to make a budget Zen 2 build from this madness with a freaking B450 motherboard!
WRITE-UP: https://level1techs.com/article/ryzen...
►► AMD BEATS EVERYTHING! (Including themselves...) AMD Ryzen 3700x & 3900X Content Creation Performance -- https://www.youtube.com/watch?v=JfEjU...
►► FASTER THAN 2080ti?! - AMD Radeon RX 5700 & 5700XT for Streamers & Content Creators (AMF UPGRADE?!) -- https://www.youtube.com/watch?v=Yi-_T...


Ted Miles 14 minutes ago
"Did the focus of the video go from 5700xt to ryzen half way through or am I blurring between the other videos and this?"

EposVox 8 minutes ago
"This video was about everything!"

 

"... On an even more disappointing note, we somehow managed to end the life of our 3900X sample at this stage of the review. We don’t recall exactly what settings were applied, but we know we hadn’t manually adjusted voltages yet. We believe after testing the 4.3 GHz overclock with auto voltage, we increased the LLC to see what impact that had on temperatures and during our first CB20 pass the system crashed and reset, and never booted up again.

The CPU now gets stuck at code 07 after microcode. We tried running the chip on a different X570 board among other tests before sadly declaring it dead. AMD says no other reviews had managed to kill their 3900X so we’re just special or unlucky, your pick. AMD has sent us a replacement but in the meantime we were unable to test the 3900X on B450 boards, or include it in the IPC test with a few cores disabled in each chiplet."
https://www.techspot.com/review/1869-amd-ryzen-3900x-ryzen-3700x/

 

We can ask him to run it again tomorrow after going through ram tuning on the stream.

 

Here are some stock numbers for the 3900x / 3700x:

Thanks to @ajc9988 for this additional chart with OC results:

" Productivity Benchmarks
There's no better place to start than Cinebench R20’s multi-threaded benchmark, and boy the Ryzen 9 3900X looks mighty. Scoring an incredible 7086 points makes it 24% faster than the Threadripper 2920X. Moreover, it decimated the Core i9-9900K by a 45% margin.

The Ryzen 7 3700X was equally impressive. The new 8-core part matched the 9900K and that meant it was 22% faster than the 2700X and 30% faster than the more expensive 9700K. These new 3rd-gen Ryzen parts are already looking like kings of productivity."
https://www.techspot.com/review/1869-amd-ryzen-3900x-ryzen-3700x/#Productivity

Similar CB20 results here:

Ryzen 3000 Review: AMD's 12-core Ryzen 9 3900X conquers all
With its ground-breaking 7nm process, AMD's Ryzen 9 3900X leave very little room for Intel's best CPUs.
Gordon Mah Ung By Gordon Mah Ung, Executive Editor, PCWorld
https://www.pcworld.com/article/3405567/ryzen-3000-review-amds-12-core-ryzen-9-3900x.html?page=2

 

Wrong chart. Needed this one:


Comparing a 5.2GHz. This would suggest that at 4.4GHz, the 3700X could further close in on the 5500 score of Talon.

Meanwhile, even though Joe was disappointed on the clocks of the chips, he said he was surprisingly having more fun than he thought overclocking it a couple times during the stream (and considering he was roughly doing the same way I overclock, even though I do things a bit different, while noting he was also on a stream, so was trying to play to an audience, not just for himself). But, he, as many enthusiasts, still want the clocks.

Flanker also jumped in chat toward the end. But, with about 300 points per 100MHz, it likely will come down to mem tuning on whether Joe can reach Talon's score at 5.2GHz on the 9900K. He'd just have to find a way for the mem and other settings to squeeze out another 100-150 points or so.

If a golden sample could reach 4.45-4.5GHz at a reasonable voltage (not seen yet considering past 4.3 the curve seems steep), I wonder how that would hit against a 9900K....

 

A good overview of the points finally being seen / recognized by the Tech Press:

AMD Finally Earns Respect From the Tech Press with Zen 2… & Navi Impresses!
Moore's Law Is Dead
Premiered 119 minutes ago
Zen 2 crushes Intel… Just like Zen 1 did, and it seems like the Techtubers get it now. I dissect Zen 2’s victory and also go into how surprisingly good Navi turned out!

Spoiler: TIMESTAMPS and Source Links:

1) Intro 0:26
2) Comparing Zen 2’s Victory to Zen 1 2:20
3) Why The Press is Finally Fed Up with Intel 4:27
4) Zen 2 Is Sandy Bridge, Zen 3 is Ivy Bridge 7:17
5) Zen 2 Conclusion 8:08
6) My Thoughts on Navi 9:42
7) What GPU’s should you buy now? 10:50
8) Overall Conclusion, and things to look for in the future 14:58

LINKS!!!!!!:
https://www.youtube.com/watch?v=z3aEv...
https://youtu.be/mW1pzcdZxKc?t=459
https://youtu.be/oDVUdpcKZMA?t=787
https://www.extremetech.com/computing...
https://www.techspot.com/article/1644...
https://www.tomshardware.com/news/int...
https://www.techpowerup.com/248407/in...
https://en.wikipedia.org/wiki/Spectre...)
https://www.tomshardware.com/news/cts...
https://www.tomsguide.com/us/zombielo...
https://www.theverge.com/2012/11/15/3...
https://youtu.be/oFziOWzVFwI?t=326
https://youtu.be/Z8X9pVIBlmY
https://www.youtube.com/watch?v=rz47W...
https://cryptoage.com/en/1663-amd-rad...
https://www.techpowerup.com/review/pc...

 

Why screw around trying to OC an 8c/16t 3700x for $329 - benchmarking on the edge of stability - when you won't be able to use that performance in day to day use - much like @Talon's OC'd 9900k on the edge benchmark?

We can get the 3900x for $499 ($170 more) - the same price as the 9900K and run it at stock speeds for better single core (CB20s 509) than 3700x (CB20s 500), and better multi-core performance out of the box (CB20m 7086) than the 9900K OC'd to the ragged edge (CB20m 5530).

Start the OC'ing with a 3900x CPU with a little more meat on it's bones, and grow the tuning from there.

Better yet, plop a 3950x down in the same AM4 socket for $250 more ($749) + higher stock clocks, and really live. By September the 5100mhz memory kit's outta be out in force, with better firmware and tuning expertise gained over the next couple of months. All for far less than the cost of a slower performing Silicon Lottery 9900k 5.1ghz CPU @ $929.

Or, wait for Threadripper 3.

Having said that, it would be fun to see a stable 4.4ghz 3700x running in daily driver mode stomping the 9900k, for "$200" less. I don't really see that happening to most 3700x's - more likely 4.3ghz stable, but maybe Silicon Lottery will have some higher clocking bins available soon.

 

you also gotta count in that zen's arch is 2yrs old with it's new SMT design vs intel's HT design been the same since sandy bridge, or possibly even first gen i7. in single threaded it is falling even with higher ipc due to overclock headroom. tbh if 4.4ghz can go with 1.25v i probably would have bought it. 1.48v is just no go.

zen3 will be optimization so maybe 3% ipc boost and another 150-200mhz boost in frequency at best. time to wait for zen 4 and tigerlake 10nm+++

 

Check out the stream tomorrow, as I think he was at 1.375v @ 4.4ghz...

Right at the end he bumped it up to 1.5v, but I don't think he needed that much voltage... he also bumped up the memory to 1.8v for 4000mhz which is probably too high as well.

He wasn't trying to finesse it he was shotgunning it after 3+ tiring hours of OC'ing, just to satisfy the peanut gallery - so he could close the stream and leave.

 

I haven't seen much of the reviews but how is the power consumption in full load with 1.375v or above oc'd?

 

I don't think he measured the power - he was just trying to get a stable run - after 3 hours of futzing around, but there are reviews with power charts showing stock power, PB2 OC, and manual 4.3mhz OC...check out the reviews I and others have posted, and check out "The Beards" streaming event video... I posted that a few pages back.

 

Well decent results, not earth shattering by any means. If the 9900k drops in price I may go that way instead, I guess I'll be playing it by ear...

 

Power Consumption & Overclocking
https://www.anandtech.com/show/14605/the-and-ryzen-3700x-3900x-review-raising-the-bar/19

Looking at the total power consumption of the new 3700X@65w, the chip is very much seemingly hitting and maintaining the 88W PPT limitations of the default settings, and we’re measuring 90W peak consumption across the package.

Edit.
[email protected] 100w with 1.4v in wprime Benchmarks.

https://www.guru3d.com/articles_pages/amd_ryzen_7_3700x_ryzen_9_3900x_review,27.html

 

Not sure where you were going with that, or what your point was... but those graphed limits are shown exceeded here:
https://www.extremetech.com/computi...0x-and-ryzen-7-3900x-reviewed-red-storm-ryzen

And, more importantly my point would be that unlike the Intel CPU's that pull back at stock settings and drop power, the Ryzen CPU's deliver continuous power:

"After a relatively short period of time (8-20 seconds, typically), the Core i9-9900K, 9700K, and 8086K will all yank back hard on the metaphorical throttle. The Ryzen 7 2700X, 3700X, and Ryzen 9 3900X do not behave in this fashion. Where the 9900K throttles back before its even halfway through a Cinebench R20 multi-core render, the Ryzen CPUs maintain full clock and power draw the whole way through."

"Prime95 29.4b8 power consumption. The Ryzen 7 3000 CPUs idle very high — much higher than their Intel equivalents, and higher than the Ryzen 7 2700X. This could improve with later UEFI revisions or might be the result of the relatively high-power PCIe 4.0-equipped chipset.

In Prime95 29.4b8, the Intel CPUs burst to between 190W – 205W for relatively short periods of time before settling back to lower wattages. This is somewhat different than their behavior in Prime95 29.8b5."

"In Cinebench R20 we still see the CPUs burst and then pull back, but they do so less aggressively and turbo clocks are held for a longer period of time. None of the Intel CPUs we tested, however, would hold its turbo clock for the entire length of time it took to finish the CB20 render in multi-threaded mode. This behavior is adjustable in UEFI — we simply used the defaults set by Asus for its motherboards.

AMD’s CPUs do not behave in this fashion, but the fact that they hit higher wattages doesn’t mean they use more total energy. The Ryzen 7 3700X and Ryzen 9 3900X both complete the CB20 render more quickly than the Core i9-9900K, which cuts down on the total power consumption. The Ryzen 7 3700X is also a significant improvement on the 2700X, drawing approximately 80 percent as much wall power in all cases."

 

See my edit. The 65W chips eat near 90W running stock clocks or 100W oc'd. I just post what I have found

In my previous post #6279 I asked what was the power consumption when oc'd and with what voltage. And I found the answer

 

To what end? What's your point?

I posted a few charts showing power draw for several apps, for Intel and Ryzen 3, check them out, they all draw more than their advertised "TDP", that's not news - you should be very familiar with that power hog behavior OC'ing Intel CPU's all these years.

" In my previous post I asked what was the power consumption when oc'd and with what voltage."

So you are trying to answer your own question? It doesn't seem to be the answer you were looking for, but it's interesting. At that link they show Overclocking and Power Consumption as independent items, there's no chart of power usage at an overclock, only performance in various tasks with results at overclocks.

 

Isn't the Hwinfo power consumption monitoring correct? I don't know. As with all hardware the power will be more or less depending on the test application. But I'm not so familiar with how correct the Hwinfo reading is for Ryzen chips.

This is for I think max stock boost on all cores for 3700X. Same chart as above but not with overclock.

 

If the 3600X drops to $200 or the 3700X hits $275-ish around Black Friday, I might jump in. My R5 2600 is still plenty good for what I need.

I don't quite know what to make of the 5700, though. It seems like results are kind of all over the place. I'll be more keen on them when non-blower AIB cards are out and the drivers have had some time to mature.

 

Again, why are you showing me this? What is your point? That the Ryzen 3 CPU's draw more real power than the TDP they advertise indicates when OC'ing outside of specifications that the TDP is based upon? You know the TDP rating isn't the power draw rating, that's not what TDP is for:

What is TDP and why should you care about it?
" TDP stands for Thermal Design Power, and is used to measure the amount of heat a component is expected to output when under load. For example, a CPU may have a TDP of 90W, and therefore is expected to output 90W worth of heat when in use."

AMD Ryzen 7 3700X & Ryzen 9 3900X review - Power Consumption and temperatures
by Hilbert Hagedoorn on: 07/07/2019 02:59 PM
https://www.guru3d.com/articles_pages/amd_ryzen_7_3700x_ryzen_9_3900x_review,7.html


That is more interesting to me, and others wondering how much power their Ryzen / Intel CPU is drawing doing various tasks.

I'm not interested in gaining 100mhz or so OC at the expense of a large power draw increase, going outside the specs that set the TDP of the CPU.

The small amount of performance increase by OC'ing manually past the PB2 auto-clock, which may not even be stable or reliable - especially if it overloads the cooling, isn't worth the trouble. There are lots of choices, why waste time creating a hot power thirsty unstable OC?

If you want more performance, get the next CPU sku up in the line-up. 3600x => 3700x => 3800x => 3900x => 3950x => Threadripper 3.

Of course that leads us to the big problem Intel has right now, there is nothing else besides the 9900k to stick in the LGA-1151 socket. Intel has nothing left in the tank while AMD Ryzen 3 has 3 CPU sku's above the 3700x 8c/16t ready for upgrading.

 

inefficient OC tbh. with intel cpu i'd never go over 1.3v just impossible to cool and not practical. these guys are using like 1.4v. with voltage like under 1.25v zen 2 could do 4.2ghz at just 1.2v which runs at around 70w?

 

Intel has a slight gaming performance lead but has lost the productivity lead all together in a big way. The 9900k can't touch he 3900x here. It no longer is where AMD is better in some workloads but just about all of them now and significantly so.

 

And I have never said this. But some complaints that the Intel chips couldn't hold within 95w. And both know why

Not sure if (all) chips can run 1.25 1.20 at 4.2GHz on all cores.

AMD Zen 2 Memory Performance Scaling with Ryzen 9 3900X Techpowerup.com
As mentioned in our main review of the Ryzen 9 3900X, heat output when overclocked is a big issue for this processor. The highest voltage that kept CPU temperatures below 95°C with our 240 mm watercooler was just 1.225 V, which is surprisingly low and limits OC potential, too.

Maximum overclock of the Ryzen 9 3900X is a nice round 4,000 MHz on both X470 and X570, so there is no difference in CPU overclocking potential between both chipsets, even with a high-powered, overvolted processor like the Ryzen 9 3900X.

AMD Ryzen 3900X & 3700X Tested on X470 Techpowerup.com

AMD Ryzen 7 3700X Techpowerup.com

AMD Ryzen 9 3900X Techpowerup.com

Of course. 50% more cores helps a lot

 

The 9900k is more power hungry for the same performance. Look at the charts showing the 3700x doing the same work, it's drawing less power. The 3900x is drawing more power but delivering much more performance from more cores / threads.

Keeping the Ryzen 3 CPU's within spec - even with PB2 enabled power draw will venture outside the spec, but not as dramatically as manual overclocking would - unless you tune the voltage down to the minimum for stable operation.

Mr. Beard wasn't doing that, he was slamming voltage down to get stability - which is counter productive on air / water cooling - he's used to LN2 saving him from being more judicious with his voltage choices.

You'd want to set up a test for both the 9900k and Ryzen 3 CPU for tasks with comparable results and see what the power draw required is in order to see the improvement in efficiency with Ryzen 3.

There are already some reviews with tests close enough to compare, but it's only the 1st day of testing, I'm sure someone will take on the efficiency comparisons for Ryzen 3 CPU's vs Intel later on, if there is enough interest.

 

Agreed, so if Intel lowers the price of the 9900K to the 3700x then the comparison changes.

 

For now, Intel have max 8 cores for mainstream. To compete they need to lower the prices. But how low, I don't know. One thing for sure... Intel need to do something if they see the sales jump down the drain But they still dominate heavly on notebooks. Maybe that compansate for the lower sales. But there is an lower limit for all.

There is an difference between brand new 7nm vs old 14++. And I didn't expect anything else. If not, something would be damn weird

 

They need to put hyperthreading back into Core i7 parts.

 

They still offer HT for the Mobile i7. I think they prefer that people buy the more expencive chips for desktops.

 

This conversation is about desktop parts. Intel is going to have to change its strategy real fast if they hope to come out of this with any kind of face. Save the Core i9 branding for HEDT and return the i7 to what it once was.

 

Intel needs to keep Hyper-Threading on everything and just segment products based on core count like AMD. Not having Hyper-Threading is literally just removing a feature after the fact.

 

Desktop and mobile chips go hand in hand. Then they need too do it for the Mobile as well. Won't happen. Remember they charge one arm and one leg for the Core i9 branding for Mobile.

 

First gen 10nm has worse transistor performance than 14nm++.

So while I get what you’re saying, the situation isn’t going to get better for Intel until 10nm+, by their own admission.

 

!.375 was not fully stable. LLC mode 2 was a little more stable, but 1.38V wasn't fully stable at 1.38V either. He tried 1.4V and it worked.

Power consumption actually looks really good comparatively, even if it seems higher than advertised, still much lower than Intel's 9th gen.

Yeah, Zen is its own thing. But with Intel, I will run up to 1.4V if cooling and thermals are good. With AMD, 1.35-1.4V is good IF AND ONLY IF your temps are good.

As Der8auer said, you will hit a wall where added voltage has zero scaling.

Flanker was able to jump in with 4.45GHz at 1.4V I believe he said, but I'd have to double check. But I'd really likely recommend less than 1.4V for a daily driver (coming from a person where 1.375V and 4.2 maxes my heat output of 68C, with very few workloads able to drive it to 72C on 1950x).

 

https://www.xanxogaming.com/reviews...ning-the-intel-core-i9-9900k/#AMD_test_system

so there maybe issue with bios revision with different version of AGESA that causing problem, especially in overclocking and boost clock. this site able to get 4.65ghz on 1.5v which i dont believe would be a sample issue. didn't think bios issue would be such a major problem in pushign clock speed.

still though, 1.5v is cancerous.

another cool thing is they did a test on SSD and looks like performance has caught up quite a bit to intel's system, which would actually be much faster than intel due to the patching.

@TANWare @Talon @hmscott