Jump to content

FMA3 bug discovered in (Ry)Zen

zMeul
5 minutes ago, Godlygamer23 said:

So you're arguing for the sake of arguing, and wasting everyone's time.

I just think he appreciate that someone wants to talk to him. Don't take that away from him.

Please avoid feeding the argumentative narcissistic academic monkey.

"the last 20 percent – going from demo to production-worthy algorithm – is both hard and is time-consuming. The last 20 percent is what separates the men from the boys" - Mobileye CEO

Link to comment
Share on other sites

Link to post
Share on other sites

2 minutes ago, valdyrgramr said:

You sourced yourself and nothing else.  That's pretty bad sourcing, so you're back to gossip posts.

no, I didn't

 

it's the same as it was with Charlie Demerjian claiming Intel paid reviewers to skew Rizen reviews, I told people to do a reality check

 

go and find the facts if you don't agree with what I say

Link to comment
Share on other sites

Link to post
Share on other sites

Lovers' spat was entertaining.

 

Is the issue with the Ryzen chips watchable via driver or BIOS? Or does it require a new CPU revision?

Cor Caeruleus Reborn v6

Spoiler

CPU: Intel - Core i7-8700K

CPU Cooler: be quiet! - PURE ROCK 
Thermal Compound: Arctic Silver - 5 High-Density Polysynthetic Silver 3.5g Thermal Paste 
Motherboard: ASRock Z370 Extreme4
Memory: G.Skill TridentZ RGB 2x8GB 3200/14
Storage: Samsung - 850 EVO-Series 500GB 2.5" Solid State Drive 
Storage: Samsung - 960 EVO 500GB M.2-2280 Solid State Drive
Storage: Western Digital - Blue 2TB 3.5" 5400RPM Internal Hard Drive
Storage: Western Digital - BLACK SERIES 3TB 3.5" 7200RPM Internal Hard Drive
Video Card: EVGA - 970 SSC ACX (1080 is in RMA)
Case: Fractal Design - Define R5 w/Window (Black) ATX Mid Tower Case
Power Supply: EVGA - SuperNOVA P2 750W with CableMod blue/black Pro Series
Optical Drive: LG - WH16NS40 Blu-Ray/DVD/CD Writer 
Operating System: Microsoft - Windows 10 Pro OEM 64-bit and Linux Mint Serena
Keyboard: Logitech - G910 Orion Spectrum RGB Wired Gaming Keyboard
Mouse: Logitech - G502 Wired Optical Mouse
Headphones: Logitech - G430 7.1 Channel  Headset
Speakers: Logitech - Z506 155W 5.1ch Speakers

 

Link to comment
Share on other sites

Link to post
Share on other sites

7 minutes ago, ARikozuM said:

Is the issue with the Ryzen chips wpatchable via driver or BIOS? Or does it require a new CPU revision?

via BIOS update

does it require a step? most likely it will since it's a HW issue - that's the whole point of stepping

Link to comment
Share on other sites

Link to post
Share on other sites

35 minutes ago, valdyrgramr said:

That's not a good argument, there.  Why don't you post the "facts" that you keep gossiping about?  Otherwise, it's just gossip.

i don't think he has "facts" 

Link to comment
Share on other sites

Link to post
Share on other sites

I'm wondering how Patrick would have reacted. Because he was one of those blowing AVX/FMA3 like there was no tomorrow.

"We also blind small animals with cosmetics.
We do not sell cosmetics. We just blind animals."

 

"Please don't mistake us for Equifax. Those fuckers are evil"

 

This PSA brought to you by Equifacks.
PMSL

Link to comment
Share on other sites

Link to post
Share on other sites

7 hours ago, djdwosk97 said:

What was nonsensical about it? When I saw some of the results they looked pretty inconsistent with other results I've seen, but I didn't bother looking further into it.

He is comparing the FX 8370 to the 2500k, and completely ignoring minimum framerates (the area in which the AMD FX series suffered the most) simply because it agreed with his point. The reason computerbase's testing methodology was nonsensical, is because they didn't include the minimum framerates in their results. AdoredTV also used a very silly method of showing the FX series improving over time, by comparing data with older titles and older GPU's against newer titles with new GPU's. He even changed the 8350 to an 8370. When you are attempting to get conclusive data in a comparison, you should not add/change variables. That's simply a poor methodology.

 

Now, I don't think he did this on purpose, as it does seem he genuinely believes the results, but it's still not an accurate representation of how this is done. If you want to prove a point, get both CPU's, do not change them, use the exact same GPU throughout all of the tests, then test both the older and new titles. From here, you can then provide the average and minimum framerates for the older titles, and compare how well the FX does in the new titles with it's average and minimum framerates. Bonus points if you then provide an over-all gaming average for your slew of tests. Perhaps AdoredTV should provide his own benchmarks in the future, to avoid mistakes like this. 

My (incomplete) memory overclocking guide: 

 

Does memory speed impact gaming performance? Click here to find out!

On 1/2/2017 at 9:32 PM, MageTank said:

Sometimes, we all need a little inspiration.

 

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

8 hours ago, i_build_nanosuits said:

...and it's been reported that AMD went as far as strongly suggesting reviewers to test gaming only at 4K...like Linus Sebastian did...why do you think he had those favors of being the only reviewer to be able to shoot inside the ryzen/vega demo machine...he did delivered ONLY 4k gaming benchmarks in his ryzen reviews...AMD also suggested a list of games specificaly to test in the ryzen reviews...you want to know the list of games? watch linus ryzen review.

Do you have sources for those (bolded) claims? I'd like to read that as well, I heard about AMD instructing reviewers on how to set up the platform for benchmarking, I also heard about Intel doing the same and instructing reviewers how to review and benchmark AMD Ryzen CPUs (obviously so it benefits them) but I haven't heard anything about what you said yet :o

 

Also, yeah, the guy's channel with AMD-fanboy thumbnails seems too off-putting and wrong to be a reliable one. I've never heard of this channel but I've seen recently a few of his videos "suggested" to me by YouTube, never even clicked on them cause those slogans on the thumbnails seemed fanboy-ish and biased...

CPU: AMD Ryzen 7 5800X3D GPU: AMD Radeon RX 6900 XT 16GB GDDR6 Motherboard: MSI PRESTIGE X570 CREATION
AIO: Corsair H150i Pro RAM: Corsair Dominator Platinum RGB 32GB 3600MHz DDR4 Case: Lian Li PC-O11 Dynamic PSU: Corsair RM850x White

Link to comment
Share on other sites

Link to post
Share on other sites

6 minutes ago, Morgan MLGman said:

Do you have sources for those (bolded) claims? I'd like to read that as well, I heard about AMD instructing reviewers on how to set up the platform for benchmarking, I also heard about Intel doing the same and instructing reviewers how to review and benchmark AMD Ryzen CPUs (obviously so it benefits them) but I haven't heard anything about what you said yet :o

Jayz2cents and Barneclese (?, can't remember the correct spelling)

"We also blind small animals with cosmetics.
We do not sell cosmetics. We just blind animals."

 

"Please don't mistake us for Equifax. Those fuckers are evil"

 

This PSA brought to you by Equifacks.
PMSL

Link to comment
Share on other sites

Link to post
Share on other sites

Just now, Dabombinable said:

Jayz2cents and Barneclese (?, can't remember the correct spelling)

Gonna look for it, thanks!

CPU: AMD Ryzen 7 5800X3D GPU: AMD Radeon RX 6900 XT 16GB GDDR6 Motherboard: MSI PRESTIGE X570 CREATION
AIO: Corsair H150i Pro RAM: Corsair Dominator Platinum RGB 32GB 3600MHz DDR4 Case: Lian Li PC-O11 Dynamic PSU: Corsair RM850x White

Link to comment
Share on other sites

Link to post
Share on other sites

8 minutes ago, Morgan MLGman said:

Gonna look for it, thanks!

I'm going to try and find the specific video-might take a while though as my internet speeds have tanked this afternoon.

Edit: I think it was this video, can't really check that easily https://www.youtube.com/watch?v=P1dhYDm7SLw&t=3825s

 

"We also blind small animals with cosmetics.
We do not sell cosmetics. We just blind animals."

 

"Please don't mistake us for Equifax. Those fuckers are evil"

 

This PSA brought to you by Equifacks.
PMSL

Link to comment
Share on other sites

Link to post
Share on other sites

On 3/16/2017 at 4:35 AM, XenosTech said:

Thing is he's going off on a senseless hate tangent as usual for anyone who like AMD. Look back a few pages and see MageTank said something about it after another user finally posted his sources for a baseless accusation

Intel 4leif bring on the 720p benchmarks

 

On 3/16/2017 at 4:56 AM, MageTank said:

He is comparing the FX 8370 to the 2500k, and completely ignoring minimum framerates (the area in which the AMD FX series suffered the most) simply because it agreed with his point. The reason computerbase's testing methodology was nonsensical, is because they didn't include the minimum framerates in their results. AdoredTV also used a very silly method of showing the FX series improving over time, by comparing data with older titles and older GPU's against newer titles with new GPU's. He even changed the 8350 to an 8370. When you are attempting to get conclusive data in a comparison, you should not add/change variables. That's simply a poor methodology.

 

Now, I don't think he did this on purpose, as it does seem he genuinely believes the results, but it's still not an accurate representation of how this is done. If you want to prove a point, get both CPU's, do not change them, use the exact same GPU throughout all of the tests, then test both the older and new titles. From here, you can then provide the average and minimum framerates for the older titles, and compare how well the FX does in the new titles with it's average and minimum framerates. Bonus points if you then provide an over-all gaming average for your slew of tests. Perhaps AdoredTV should provide his own benchmarks in the future, to avoid mistakes like this. 

Yeah this is a valid point. He makes mistakes to like every Youtuber does. Doesn't make him a shit channel. I was also led to believe the 8350 improved when it likely didn't looking at the big picture. And the power consumption is still bad...

 

The point about Ryzen still stands, it's an awesome CPU and a beast in productivity with alot of potential left unused in the gaming sector. For now the 7700k remains king in gaming, although it does not really matter if you have a GTX 1070 or below...

 

\\ QUIET AUDIO WORKSTATION //

5960X 3.7GHz @ 0.983V / ASUS X99-A USB3.1      

32 GB G.Skill Ripjaws 4 & 2667MHz @ 1.2V

AMD R9 Fury X

256GB SM961 + 1TB Samsung 850 Evo  

Cooler Master Silencio 652S (soon Calyos NSG S0 ^^)              

Noctua NH-D15 / 3x NF-S12A                 

Seasonic PRIME Titanium 750W        

Logitech G810 Orion Spectrum / Logitech G900

2x Samsung S24E650BW 16:10  / Adam A7X / Fractal Axe Fx 2 Mark I

Windows 7 Ultimate

 

4K GAMING/EMULATION RIG

Xeon X5670 4.2Ghz (200BCLK) @ ~1.38V / Asus P6X58D Premium

12GB Corsair Vengeance 1600Mhz

Gainward GTX 1080 Golden Sample

Intel 535 Series 240 GB + San Disk SSD Plus 512GB

Corsair Crystal 570X

Noctua NH-S12 

Be Quiet Dark Rock 11 650W

Logitech K830

Xbox One Wireless Controller

Logitech Z623 Speakers/Subwoofer

Windows 10 Pro

Link to comment
Share on other sites

Link to post
Share on other sites

29 minutes ago, Dabombinable said:

I'm wondering how Patrick would have reacted. Because he was one of those blowing AVX/FMA3 like there was no tomorrow.

Yea only problem with that is not everything can or does benefit from AVX/FMA3, he was implying everything would benefit from it but that just isn't true. Then there is the whole thread synchronization issue, if speeding a few things up can be done but you then have to delay wait for other threads then there is no gain.

 

Games are actually rather complex and it is never as simple as saying use X or Y instruction set since it gives improved performance, a true statement in isolation but maybe not overall.

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, Dabombinable said:

Jayz2cents and Barneclese (?, can't remember the correct spelling)

 

1 hour ago, Morgan MLGman said:

Gonna look for it, thanks!

exactly, Jayztwocents spilled it out in techtalk which i always listenned too (show is over now btw)...it's a 1.5hr long weekly show and i'm not sure about whicih episode it was and i will not go look back but jay said it was the reason he did not made the usual ryzen review that people were expecting from him instead he just made a watercooled ryzen build...

| CPU: Core i7-8700K @ 4.89ghz - 1.21v  Motherboard: Asus ROG STRIX Z370-E GAMING  CPU Cooler: Corsair H100i V2 |
| GPU: MSI RTX 3080Ti Ventus 3X OC  RAM: 32GB T-Force Delta RGB 3066mhz |
| Displays: Acer Predator XB270HU 1440p Gsync 144hz IPS Gaming monitor | Oculus Quest 2 VR

Link to comment
Share on other sites

Link to post
Share on other sites

4 hours ago, Dabombinable said:

I'm wondering how Patrick would have reacted. Because he was one of those blowing AVX/FMA3 like there was no tomorrow.

it would probably be something along the lines of:

"something something blue dragon something something AVX is the future something something IPC can't exceed 1 something something .... "

Please avoid feeding the argumentative narcissistic academic monkey.

"the last 20 percent – going from demo to production-worthy algorithm – is both hard and is time-consuming. The last 20 percent is what separates the men from the boys" - Mobileye CEO

Link to comment
Share on other sites

Link to post
Share on other sites

I don't know if this has been already mentioned here, but this bug is not reproducable on Linux. Only on windows and SMT enabled.

 

German source: https://www.golem.de/news/fma3-instruktion-windows-smt-bug-laesst-ryzen-systeme-abstuerzen-1703-126770.html

Quote

Interestingly, however, the error does not occur when the simultaneous multithreading (SMT) of the CPU is turned off. This does not support all mainboards, which is why Yee can not adjust this behavior. However, a user of the Hwbot forum also confirms that the crash does not occur when the SMT is switched off.

The error is probably also limited to Windows. On Linux, the application does not cause any crashes, regardless of SMT usage. Yee and other users also acknowledge that Linux is not affected. The execution of the Windows binary files under Linux is also fault-free with the help of Windows's Windows API replica.

(Google translator)

Link to comment
Share on other sites

Link to post
Share on other sites

more info on this from PCPer https://www.youtube.com/watch?v=dDrOQrjJTSs

it seems that the FMA3 set isn't actually bugged, but it seems it doesn't have enough power delivered to it as 25% of the people who OCed their CPU were able to ran the test successfully

 

sources inside AMD have said that it's a known issue and a BIOS update is indeed planned

Link to comment
Share on other sites

Link to post
Share on other sites

On ‎15‎/‎03‎/‎2017 at 6:25 PM, zMeul said:

-snip-

At least the AM4 motherboards VRMs aren't blowing up like with X99s launch. xD

        Pixelbook Go i5 Pixel 4 XL 

  

                                     

 

 

                                                                           

                                                                              

 

 

 

 

Link to comment
Share on other sites

Link to post
Share on other sites

On 16/03/2017 at 6:29 PM, leadeater said:

Yea only problem with that is not everything can or does benefit from AVX/FMA3, he was implying everything would benefit from it but that just isn't true. Then there is the whole thread synchronization issue, if speeding a few things up can be done but you then have to delay wait for other threads then there is no gain.

 

Games are actually rather complex and it is never as simple as saying use X or Y instruction set since it gives improved performance, a true statement in isolation but maybe not overall.

Well said! If I may elaborate, consider these:

 

1) In-house game engine with self-created libraries. In this scenario, if games have already been built on this engine over many years, there's a very hefty cost/benefit ratio to implement new feature sets within pre-existing ones. Both WoW and Guild Wars suffer from such scenarios, as they are so entrenched within pre-existing code that making any updates becomes an immense undertaking. A complete re-code has other followups, as there'd be re-training to update ongoing games or rework currently developing projects.

2) In-house game engine with external dependencies. Same as per (1), except that now developers don't really want to update dependencies, preferring to use deprecated code rather than risk breaking something and creating a huge mess. It takes away control from the developers when it comes to development work too.

3) External game engine. Relevant to (2), but now the engine developers (Unreal, Unity, etc) now have to be careful not to make too many changes and create an impossible learning curve for game developers. If a new game engine becomes impossible to learn and deploy within a reasonable amount of time, then it'll just increase the chances of developers preferring to rely on deprecated code just to stay sustainable.

 

At its very core, I think software and hardware development work is about business sustainability. It can't be too costly, and this isn't just about money! It's about time: CPU processing, GPU processing, learning new processes, development and experimentation, as well as the process of continual product support over the years.

So, yes, brute forcing an algorithm through pure dual core processing speeds isn't optimal. But for developers plagued by support tickets, bills to pay, loans to manage, and families to feed, why spend those extra days trying to figure out how to implement a new API or library or instruction set?

In a similar vein, I'm guessing AMD very likely chose to focus their limited resources on the core unchangeable features of Ryzen, and then deploying other changeable features "as is", since they can be patched later.

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now


×