AMD Ryzen 7 2700X User Publishes Benchmarks and Overclocking Results | Hacker News Reader

AMD Ryzen 7 2700X User Publishes Benchmarks and Overclocking Results

wccftech.com

115 pointsO_H_E83 comments

Lovely news! This is my next CPU after many years of my trusty FX8350@4.8 GHz. Last time I waited for the second revision of the architecture to avoid the usual launch hurdles. Christmas is apparently coming in April this year.

Same here. As a web developer and gamer it's hard to ignore the Intel lineup, but I'm a big fan of supporting the under-dog, and it makes it easy when there's so much value packed in.

I assume that motherboards with new chipsets will be released as well, but I've heard less about them.

The Ryzen 2 lineup should work on the same AM4 socket, AMD claims that the newer X470 chipset will give better performance.

FX-6300 over here. Been waiting to upgrade for a while and looking forward one of these new AMD cpus.

As someone who purchased (early adopted) the Ryzen 5 2400g APU a couple of days after release, I regret every bit of it. We still don't have any motherboards that support it well. I've heard the ASUS boards do okay but they have to be flashed to a new BIOS version (with a different processor in the socket). I'm glad that folks like the hype, but the reality is just terrible.

what would a good motherboard support bring ? (open question)

Processor support? Voltages set reasonably. Just a decent burn in period before sending it out the door. Currently I'm running an ASRock B350M Pro4... and I can't keep that thing running more than 30 minutes. It's not high load that causes the freezing. It's just terrible on every front.

erf, damning indeed

The first generation Ryzen 7 rollout was a bit of a mess as well. What with the (understandably) immature BIOSes, and tons of people RMAing CPUs to be replaced due to the segfaulting issue.

AMD seem gaffe-prone. See also the pre-launch hype vs reality of Vega in the gaming space.

And there is still this: https://bugzilla.kernel.org/show_bug.cgi?id=196683

Fist gen ryzen desktop had the same sot of teething troubles ever CPU has tbh, but they were somewhat moe publicised and a little bigger (not huge but bigger) due to the cleanroom design.

I got it running on asrock ab350 gaming-itx/ac issues found so far:

- No dual screen support (the board has 2xHDMI 1.4). The fix is in firmware 4.51A which is still in beta

- When installing debian needed to use vga=771 in the console

- X11 would freeze a few times but with Linux 4.15.0-2-amd64 that issue seem to have been fixed

That's why you typically wait to be 'second best' when new tech hits the market. In the current climate 'early adopter' should actually be interpreted as 'beta tester'.

I built a standard (non gfx) Ryzen to replace my old back up box, and it's been 99% fine. The only issue is the slow POST time, but that's an MSI thing and not a Ryzen thing.

The rollout of new mainboards and Zen+(non gfx integrated) CPU's should be a lot smoother than their initial effort since they've all had a year to learn from their mistakes.

Other than the BIOS issue I have had no problems with the Ryzen 5 2400G. I have an ASRock board. I tried running Warframe on only the APU but I think I need more memory which is very expensive so I put in an GTX 650.

If memory prices go down I should be able to use it with only the processor.

I did 16gigs of 3200. Be careful as the processor will only support 2400 or 2600Mhz (I can't remember off the top of my head) I don't mind the actual layout or operation of the BIOS. I did have to upgrade it to recognize my M2 NVMe. I've just been trying anything I can to get the thing stable.... to no avail.

> I regret every bit of it.

Why make a statement without even one item of support as to why you regret it?

The ASUS board problem sounds inconvenient, to say the least, but the way your phrased it implies it doesn't effect you in particular.

Comments on a popular online electronics retailer's website seem to be positive for this model.

I'm pretty sure he regretted it because "[w]e still don't have any motherboards that support it well."

What does that mean exactly?

No, I've heard good things about the ASUS. The ASRock (which is what I bought) is absolutely terrible. I regret buying the ASRock because I now have a computer that freezes constantly. It's not related to processor load or heat. It simply just doesn't work well at all. Sometimes an hour, sometimes 5 minutes....

I'm hoping I'm phrasing this well enough.

FWIW I am running a two week old Ryzen 5 2400G and ASRock Fatal1ty X370 Gaming-ITX/ac setup with rock solid stability under Windows 10. I did end up having to run my 3200MHz memory at 2933Mhz to avoid some issues. This is the max memory speed the CPU supports anyways so I don't consider it a large loss. The ASRock published memory QVL agrees with my finding that 3200MHz isn't stable with the specific memory I'm using. See https://www.asrock.com/MB/AMD/Fatal1ty%20X370%20Gaming-ITXac...

What ASRock model are you using? Are you sure the culprit is the CPU or motherboard and not another component?

I'm running the micro-atx AB350M Pro4 AM4. One can never be TOO sure that it's not something else, but I've run it on one stick of ram, with no other peripherals (removed WiFi PCIe card) and the problem persists. Still, the single stick of ram is the best it's ever run. I'd love to run at 2933, but it "seems" to freeze even more at those speeds. I hate that this is all non-scientific anecdotal evidence. That seems to be right in line with everything else over at the ASRock forums.

As an owner of an x360 Raven Ridge Envy, I'm also having similar issues. GPU Drivers on Raven Ridge seem to be borked at the moment.

I'm running Win10, default drivers (which seem to be 17.7 GPU drivers). HP Envy x360 laptop CANNOT upgrade to more recent drivers.

Have you tried updating to 18.3 or later GPU drivers?? Raven Ridge is allegedly all the same die, so I wouldn't be surprised if the iGPU drivers between both of our systems have a similar issue.

But yeah, its looking to me like Raven Ridge (the iGPU parts) still have some warts. The pure CPU Ryzen parts had warts too when they were released, but they were fixed much more quickly.

I'll give updating the GPU drivers a shot and report back this evening. It's now become my "I'll try anything" setup.

Yeah, its a shame too. As long as my laptop doesn't crash, its actually kind of awesome.

But crashing makes it unable to do any kind of "serious" work. Nothing breaks a workflow like a forced unexpected reboot.

This "crashing" issue seems to be isolated to Raven Ridge. I've seen a lot of complaints about it here and there online. There was an obscure crash on Pinnacle Ridge CPUs (1800x, Threadripper, and EPYC) on Linux, but it was allegedly fixed months ago.

So based on my experience and research, it seems like Raven Ridge needs to be avoided. Its drivers just aren't good enough for prime time yet. Let me know if you're able to successfully update to something newer.

Oddly, wiping the thing out and installing with a brand new download of Windows 10 from MS's site has done wonders. My old install media was pretty old. I haven't installed a single external driver for anything this time around and it hasn't crashed. My Vega video driver was even recognized. I'm gonna let it run for a day or so before declaring any victory. Fun side note, I JUST placed an order for an ASUS mobo to replace this one.

Updating BIOS is a common thing for newly released CPUs. Things like that are expected if you're building the PC yourself.

But needing a different chip in the socket? That’s definitely not common and for many not even possible.

Sure, was the same thing for my AMD FX cpu.

Edit: I bought a cheaper cpu on Amazon and after the BIOS update for the new CPU I sent it back.

AMD offered a kit to update the BIOS, basically they sent a previous generation CPU to flash the BIOS.

On the other hand, with new Intel CPUs you would need new mainboards... ^__^;

Yes, and kudos to AMD for even doing it. But, it's a two week from acceptance into the program to delivery (in some cases - a week in others).

I've built a lot of machines, and I've never had to update BIOS just to get a standard CPU working.

You tend to if you use a motherboard with an older chipset with a newer generation CPU using the same socket (and it hasn't already been updated at the factory).

Intel example: Using a 7th generation Kaby Lake CPU on a 6th generation Skylake chipset motherboard. Example of required BIOS update: version 1.F https://www.msi.com/Motherboard/support/Z170-A-PRO

I clearly remember my i5 4690 wasn't supported by z87 motherboards unless you flashed with another CPU.

As I was building a new rig, I just ordered a z97 instead.

Also yeah, it was my fault for not properly checking before buying 4690+z87 combo, but not that weird...

I got a Ryzen 7 1800X machine from Dell and regret it. It crashes with different BSOD error codes from time to time. A great benchmark is useless if it can't be used smoothly.

I wouldn't blame the CPU as much as I would blame Dell. I have had a Threadripper 1950x (2 Ryzen 7 1800X on the same chip IIRC) since launch day and have not had any stability issues (unless I try to overclock it over 4GHZ - but that's my own fault/doing.)

The launch of Ryzen G was undoubtedly botched, but AMD did offer a CPU loan program for anyone who needed it:

https://www.pcgamer.com/amd-will-lend-you-an-older-processor...

Isn't that for updating the bios on a board that won't boot with a Ryzen in the socket? It's a 10 day loan.

I bought a 2400G with a motherboard[0] a day after release. It surprised me that the board came with a compatible BIOS out of the box. (I was thinking that I might have to take my 1800X out of my main rig to flash it.) I built the machine for a friend, and he's been doing some light gaming on it (with a discrete GPU). I ask him every now and then how it's doing, and he says there are no problems.

[0]: https://www.newegg.com/Product/Product.aspx?Item=N82E1681315...

Beating an 8700K and 7820x, not bad at all. Nice to see AMD giving Intel some competition again. Hopefully this leads intel to drop prices,

The 8700K's main draw over the Ryzens is single-core performance though.

True, but it looks like these 2700X will come out under the 8700K's price point.

So if a Ryzen7 2700X has better multithreaded (aka: compile times, blender rendering, etc. etc.) speeds than something spec'd ~$50 to ~$100 more than it, that's a winner by any stretch of the imagination.

Single-core performance is most important in Javascript and Video Games. But most production tasks these days are multithreaded.

Admittedly niche, but my main draw for the 8700K was Hackintosh compatibility.

Nice incremental update, but since I already have a Ryzen 7 1700 on an Asrock X370 Taichi and it's been wonderful, I'll take a pass. If I hadn't already upgraded from my FX-6100, though, it would be worth looking into this one. As it stands, already having the 1700, I'd move up to a Threadripper if I could afford to upgrade, since a new motherboard would be needed anyway.

> since a new motherboard would be needed anyway.

Minor correction: The Ryzen 2000 series uses the same socket, so you can use the same motherboard, although AMD claims that the newer X470 chipset will give you better performance. Either way, I agree with your assessment: It doesn't seem worth it to upgrade if you already have a Ryzen 7.

The x470 and 370 chipsets are identical it’s a rebrand. They changed some of the power delivery requirements to match the higher TDP of some of CPUs and apparently not all previously made boards were upto snuff or at least that is the party line.

Running an i7-4790K here, and still torn on upgrading, would like the extra threads and going more than 32gb ram, but not sure it's worth it or not still.

> It also comes with a clear fan, for better RGB light illumination than its predecessor which only had an RGB ring.

How did the RGB naming trend get started?

Rgb leds and colorful illuminated gaming pc's both already existed for decades. But only recently I saw the name RGB pop up in motherboards, keyboards, etc... as if it's something new

I recently built a new PC so I had to wade into this space again. The main difference I believe is that many components themselves now have their own LED lighting.

Before, it was primarily LED fans or light strips. Now, motherboards have their own LED line traces. Ram has it's own LED lights. There is also support to control the colors of these components from software.

It’s also essentially impossible right now to buy a motherboard that isn’t tarted up with a bunch of stupid LED nonsense. (Unless you buy something low-end or server-grade.)

Yeah it really ruins it honestly, back when I built machines I always went with black aluminium Lian li cases with no windows since they had a nice professional look. I just hate useless lights on anything.

You know you can turn it off right?

On my board, I'd have to install some Windows application to do so. No thanks, I'll just close them up inside my windowless case.

Not true. Asus and some other brands have Workstation class motherboards without the LEDs.

https://www.asus.com/Motherboards/Workstation-Products/

Fine if you want old stuff. None of those boards support Coffee Lake chips though.

I just wish the lighting told me something useful. If the CPU fan changed color to indicate temperature or CPU load, that would be really cool. Likewise, if RAM lighting indicated the rate of data transfer.

That would be pretty awesome. But all this fluff just to make PC more pretty seems pointless to somehow.

(Don't get me wrong, if you like that sort of thing, by all means, enjoy it. My sister's boyfriend has a gaming PC with a big glass window in the case so one can see all the pretty lights quietly shifting around. It is pretty, and I bet it's boatloads of fun to look at when tripping.)

>If the CPU fan changed color to indicate temperature or CPU load

With Asus's "Aura" application that comes with their new boards you can have the LEDs change color with CPU temp. The default was to set everything to green/yellow/red depending on range but it might be capable of something more complex. The lights are addressable so your CPU idea might be doable I haven't really played with it yet.

That would be cool. I am a little OCD when it comes to monitoring. I can be perfectly patient with a slow computer as long as I know why that machine is slow. But if a computer goes slow on me and does not offer me any way to find out why it is so slow, that drives me mad.

So if the lighting in the PC reflected the components busy-ness, that would be really cool. "Why is my PC so slow today?" - "Aaah, the hard drive is shining white - I/O overload!"

Anyway, my PC has a window-less case, so all that beauty would be lost like tears in the rain...

That would be really cool but the wiring might get a bit too complicated. Also the lighting is slapped on to readymade parts, so it's not like you can do alterations to the silicon that would be necessary for those features. With that said, since we're seeing programmatically controlled RGB lighting getting released, maybe someone could write a lightweight daemon to control it based on those parameters.

> I just wish the lighting told me something useful.

There's a cross-platform, open source project that does this[1]. You can even script it and bind colors and effects not just to CPU load but to any variable (I don't know - maybe requests being served per second)

1. https://github.com/dcerisano/RGB-LED-Motherboard-Header-Driv...

That is really, really cool!

My PC case does not have a window, so I will never know the difference. But it is very cool one can do that!

That naming has been used for a few years already. I'm not sure how it started.

For example, this Linus Tech Tips video about the first RGB backlit mechanical keyboard from 4 years ago uses the term:

https://www.youtube.com/watch?v=y-GjMH8kQlc

I think a part of it is the Instagram/Twitch mentality where publicly performing or sharing something plays into things now.

Easier to take photos of your computer and share them if they look like a cryogenic chamber. :)

In addition, a lot of companies are going "full stack" for computers, and a system-wide RGB integration/syncing for Company X's components (RAM, motherboard, graphics card, case, fan controller, fans, mouse, keyboard, mousemat(!)) is a simple way to lock in customers.

EDIT: You could also argue that we're at an inflection point where videogames and tech nerdiness are normalized to the point where computer cases, like TVs once were, are now considered furniture and not some heinous eyesore you're forced to have around.

My hypothesis is, most assembling is now done by young gamers, as majority other users have migrated to laptop/tablet. So flashing PCs is now major chunk of the market.

It's become conventional to use it to refer to a light with customizable RGB parameters, versus a single color. For example both my desktop keyboard and mouse are "RGB" and they have various color modes and even animation settings.

Looks good. I am going to wait to see Intel's response, but this may finally be worth upgrading from an overclocked 2600k mostly used for gaming.

I wish AMD will make newer generations of their desktop CPUs chipsetless just as their motherboard cores. For Intel, going for a separate chipset is a requirement dictated by their design process and big die sizes, for AMD, it is just a way to make an additional $10-$20 per CPU sale.

Chipsetless setup should allow for smaller motherboards with better POL setup.

So what's the best upgrade path from 1700X, 2700 or 2700X? 2700X sounds more like an upgrade from 1800X.

Why do you want to upgrade the 1700X? What does the 2000 line bring other than a 15% (being very optimistic) performance gain?

Personally, because my CPU has a hardware bug[1], which is now worked around by disabling package C6 state, and I already went through RMA for the segfault bug, so I'd rather now get a newer CPU to begin with.

[1]: https://bugzilla.kernel.org/show_bug.cgi?id=196683

I think I've read locks/hangs' reports even with newer APUs, and at this moment I'm not sure if disabling C6 state on package (or was it core?) is an effective workaround.

Or the suggestion to slightly increase voltage or a light overclock.

I wanted to upgrade to Ryzen but I don't want to spend time to troubleshoot the HW. Seems like it's only triggered on a BSD/Linux system, don't know if they were joking when they said that Windos is never that idle...

> I think I've read locks/hangs' reports even with newer APUs, and at this moment I'm not sure if disabling C6 state on package (or was it core?) is an effective workaround.

For me it is. It's triggered by the firmware setting:

    Advanced > AMD CBS > Zen Common Options > Power Supply Idle Control = "Common current idle" (instead of auto)

That disables package C6 state for me (checked with zenstates.py), and freezes and random reboots disappear after that:

    C6 State - Package - Disabled
    C6 State - Core - Enabled

On auto setting for the above, I get:

    C6 State - Package - Enabled
    C6 State - Core - Enabled

And it highly annoyingly randomly freezes and reboots in such case.

Another workaround is in the linked thread, to compile the kernel with special rcu options, I tried that, and it works, but it's not really a good method.

> don't know if they were joking when they said that Windows is never that idle...

I don't think they are joking.

I've read of people that experienced hangs even with the RCU workaround.

About the C6 state is good to know, but the lack of some kind of official confirmation from AMD is kinda disturbing as it looks like the change to the BIOS came from them.

I'm really torn...

Edit: disturbing is not the right word, disappointing is better

Yeah, lack of communication on these bugs from AMD is really annoying, especially since there is no way to know whether they are fixed in newer CPUs already or not.

Clickbait, overclocked Ryzen 7 2700X scores the same as i7 7820X.

It's a bit different story

Sorry, please don't judge the site. This is my own title. Can you please elaborate on how is it a different story?

Is it overclocked though? The article says it was overclocked to 4.3GHz. But that CPU has a turbo clock of 4.2GHz [1]. So maybe a 0.1 GHz overclock at most? Or none at all? That Intel CPU has a turbo clock of 4.5GHz, so AMD beats Intel at a lower clock speed? But then I read that turbo clock only affects the two most used cores... I guess it's pretty hard to compare lol

1. https://overclock3d.net/news/cpu_mainboard/ryzen_2700x_spott...

That's the maximum turbo clock, which only kicks in when one or two cores are in use. Getting all cores to that speed does count as an overclock.

The max turbo (xfr) is 4.35ghz. But it rarely hits that

For about $100 cheaper..

The big news here is Intel not having a monopoly on desktop CPUs anymore. I personally remember Intel being the only serious CPU choice for building a desktop since at least 2015. Now a lot of low-medium high desktop builds are using Ryzen instead for the perf/$.

Thats because AMD basically bet the farm on the GPU replacing the FPU for doing floating point math, but never got any real traction for it on the OS level.

I'd buy that argument if they had actually showed any investment in helping that assumption come to pass. Instead, their ecosystem is a barren wasteland compared to what Nvidia has built.

> For about $100 cheaper..

More than that. An AM4 motherboard is way cheaper than Intel Extreme FCLGA2066 systems.

By the end of the day, you're $200 cheaper once you account for a completed build (motherboard, RAM, CPU) compared to a FCLGA2066 + i7-7820X system.

Really, the i7-7820X competes against Threadripper 1900X. Quad-channel memory and all that good stuff. The REAL competitor to the Ryzen 2700X is the Intel i7-8700k, which has outstanding price/performance and relatively cheap motherboards too.

Hopefully we'll see more reviewers compare these two chips. Ryzen 2700X vs i7-8700k is the real battle.

Reply on news.ycombinator.com