Overclock.net - An Overclocking Community

Overclock.net - An Overclocking Community (https://www.overclock.net/forum/)
-   AMD CPUs (https://www.overclock.net/forum/10-amd-cpus/)
-   -   CES19: AMD Zen 2 results.... Your Opinion? (https://www.overclock.net/forum/10-amd-cpus/1717918-ces19-amd-zen-2-results-your-opinion.html)

Storx 01-10-2019 07:19 AM

CES19: AMD Zen 2 results.... Your Opinion?
 
Hello All,

I wanted to get some input from the masses on what you think to come with the Zen 2 cpu's coming out mid 2019 now... Just going off what we were shown it appears there is room for a 2nd chiplet, or possible room for the GPU on those chips, but the cinebench scores were a little underwhelming in my opinion. Whats your take on all of this?


Just going off facts we already know...

Intel 9900K stock usually can acheive the score they shown with stock boost of 5/5/4.8/4.8/4.7/4.7/4.7/4.7Ghz
but...
We also know people have gotten similar or slightly higher scores in cinebench with extreme overclocks of the Ryzen 2700's to 4.35Ghz

So this leads me to think that the next Zen 2 cpu is only going to turbo up to 4.4 maybe 4.5Ghz, which ultimately is useful in production with lots of cores, but modern games are still designed around frequency as king in performance really...., heck the boost very well could only go as high as 4.3Ghz, but increase in IPC from the new I/O could make up some that held back the 2700's.

Even if they do come out with a 16 core version of the CPU later, i feel like the lack of clock is going to set them back in the gaming world unless your looking to use the CPU for both gaming and production which is a small fraction of the overall market. I do think the drop in power draw is huge on controlling temps and stability, especially on air cooling, but i feel like this is a wash if your already invested in high end cooling solutions, because the drop in power usage is just a few cents less of power your going to draw per year. I was really looking for better results from the released info, as i was waiting on ordering my next CPU choice based on this release, but feel like they let us down ultimately on the cpu, they very well may be able to overclock to higher, but then there is the issues with less overclockability with the smaller 7nm chip, because they can only take so much before damaging the chip.

speed_demon 01-10-2019 07:23 AM

It's too early to tell. We had less of a tour-de-force and more of a small taste of what's to come. Just the bait to keep us on the line.

ENTERPRISE 01-10-2019 07:28 AM

At this point in time we can only speculate, we have some information based on rumors/reliable sources but not anything massively concrete on which to formulate a decision.

white owl 01-10-2019 07:36 AM

Underwhelming? You speculation was that AMD beat a 5ghz/4.7Ghz CPU with a 4.4Ghz CPU while using 30% less power. This just drives home the thought that AMD can beat Intel core for core with lower clocks and less power and it's still not good enough lol
Who said that there are issues overclocking on 7nm? TSMC was pretty confident they could do 5Ghz.
When we went from 22nm to 14nm on Intel the safe voltage went up to 1.4v FWIW

The games don't just care about clockspeed if the IPC is much better. Games want strong single thread but now they're starting to favor more cores as well. Say you have two single core CPUs, one at 4Ghz and one at 5Ghz yet they score exactly the same in Cinebench. Games aren't going to see much difference there, sure more clock speed is always better on top of good IPC but if you can get more instructions across with less cycles then IMO that's the better arch.

Keep in mind we were watching a 6 month early ES chip which are ALWAYS clocked lower or equal to the final thing. Sometimes they don't even have boost/turbo.

Storx 01-10-2019 08:11 AM

Quote:

Originally Posted by white owl (Post 27798870)
Underwhelming? You speculation was that AMD beat a 5ghz/4.7Ghz CPU with a 4.4Ghz CPU while using 30% less power. This just drives home the thought that AMD can beat Intel core for core with lower clocks and less power and it's still not good enough lol
Who said that there are issues overclocking on 7nm? TSMC was pretty confident they could do 5Ghz.
When we went from 22nm to 14nm on Intel the safe voltage went up to 1.4v FWIW

The games don't just care about clockspeed if the IPC is much better. Games want strong single thread but now they're starting to favor more cores as well. Say you have two single core CPUs, one at 4Ghz and one at 5Ghz yet they score exactly the same in Cinebench. Games aren't going to see much difference there, sure more clock speed is always better on top of good IPC but if you can get more instructions across with less cycles then IMO that's the better arch.

Keep in mind we were watching a 6 month early ES chip which are ALWAYS clocked lower or equal to the final thing. Sometimes they don't even have boost/turbo.

I honestly get what your saying, its a great improvement if a 4.4Ghz chip can match the performance of a 5.0Ghz chip on cinebench, which will be better in almost all cases unless your a gamer tho..

As a streamer, i have multiple friends with pretty high-end rigs, my one friend has a high-end custom watercoooled AMD 2700X setup which is overclocked to 4.35Ghz, which regularly gets cinebench scores of 2060-2080's, but he had to turn it down to 4.3Ghz due to some games he plays are unstable at the 4.35Ghz, but still at 4.3Ghz he can pull out a 2050-2060 score easily. His roommate who also is a streamer recently build a new rig 9900k w/ AIO only and just leaves his on 5.0 all core lock and lastnight after CES i asked him to benchmark his while we talked about the release and his posted up 2145 score, but afterwards we reset it back to default bios settings and it pulled a 2022 score.

So just comparing the 2 rigs, playing the same exact game together running the AMD rig overclocked to 4.3Ghz and running the Intel rig at stock settings and both have RTX 2080ti's, the Intel rig is pulling about 10-18 fps more average in the games, but bumping the intel cpu to 5.0 locked just widens the gap more..

Just seems the higher clocks are always going to win over in gaming when using the same GPU, even if the lower clocked chip is equal or higher in IPC

Dogzilla07 01-10-2019 08:37 AM

That ES Ryzen was running at up to 4.6GHz vs the 9900k. (No clue if it was PBO boost, or if they OC-ed all cores to 4.6GHz)

rdr09 01-10-2019 08:40 AM

Quote:

Originally Posted by Dogzilla07 (Post 27798998)
The ES Ryzen 3 was running at up to 4.6GHz vs that 9900k. (No clue if it was PBO boost, or if they OC-ed all cores to 4.6GHz)

And to speculate further, they used the 65W cpu bcos my 65W R7 2700 Oc'ed to 4GHz uses around 135W at load in Cine15 (just based on HWINFO).

The Pook 01-10-2019 09:00 AM

Quote:

Originally Posted by white owl (Post 27798870)
Underwhelming? You speculation was that AMD beat a 5ghz/4.7Ghz CPU with a 4.4Ghz CPU while using 30% less power.


Not sure you can consider an unannounced un-spec'd CPU overclocked or stock or underclocked :p

Was certainly impressive but I'm interested in what the off-set IOand chipplet/unused PCB space might mean. 16 core rumors might not be far-fetched. :specool:


Quote:

Originally Posted by Dogzilla07 (Post 27798998)
That ES Ryzen was running at up to 4.6GHz vs the 9900k. (No clue if it was PBO boost, or if they OC-ed all cores to 4.6GHz)


How does anyone other than AMD know that? :o

white owl 01-10-2019 09:15 AM

Quote:

Originally Posted by Storx (Post 27798932)
Just seems the higher clocks are always going to win over in gaming when using the same GPU, even if the lower clocked chip is equal or higher in IPC

I think you're misunderstanding this part. Games don't scale perfectly across many cores, that much is true but games and other applications don't care about clockspeed alone. Look at the FX chips, they did really well in multi but despite their very high clockspeeds Intel's IPC advantage always won in games even with lower clocks.
Games don't care about clockspeed, they care about single threaded performance. Now we're getting to a point where that's true but games can also leverage the core count up to a point.
IPC = Instructions per clock
If you have low IPC you can speed up the clock to get the same amount of instructions across in the same time frame. If you have IPC you can lower the speed and have the same work done in the same time.
Games don't care one way or another as long as the same work is being done in the same amount of time on a single thread. The FX series is living proof that games need more than just speed.

Dogzilla07 01-10-2019 09:47 AM

Quote:

Originally Posted by The Pook (Post 27799052)
Not sure you can consider an unannounced un-spec'd CPU overclocked or stock or underclocked :p

How does anyone other than AMD know that? :o

AMD said it to attendees at the conference.

https://uploads.disquscdn.com/images...7b087fdbef.png

courtesy of Videocardz comment section.

Oh also in the Anandtech article Ian surmised that the new ryzen chip was running 75W, and that the 9900k was running 125W, so that gives a really nice baseline to compare IPC.

I mentioned in the other 9900k post here on the forums, that this leaves AMD room for 105w clocks, and 135w clocks. Now hopefully 2-core 5.0GHz PBO fits into 105w, and all core 5.0GHz fits into 135W
though a full core 5.0GHz is really a best binned best case scenario, and we don't know if AMD is gonna let us have these dies outside Epyc 2 and Threadripper 3 xD or if papa Physics, mother Yield and uncle Defect are gonna let us have those either :P

constructorx 01-10-2019 12:35 PM

I think what we saw was a Ryzen 5 beating a i9 9900k and the Ryzen 7 will be 16 core.

Speculation of course.

martinhal 01-10-2019 02:19 PM

At least we can expect 9900k performance on our current AMD boards in six months. My concern is pricing. Is AMD going to be agressive to gain market share ? If they edge out Intel just a little bit in gaming they may become the go to CPU for GPU benchmarks. Once that happens they don't need to price too low as the product sells its self. That could be the start of a price war , who knows.

elguero 01-11-2019 01:07 AM

Quote:

Originally Posted by constructorx (Post 27799420)
I think what we saw was a Ryzen 5 beating a i9 9900k and the Ryzen 7 will be 16 core.

Speculation of course.

I also think the processor we so, was a Ryzen 5, but Ryzen 7 probably will be a 12 core, I don´t see AMD releasing 16 core just yet, maybe later on the year.

Don't get me wrong, AMD could/can/will release a 2 chiplet 16 core any time they want, but I just don´t see them doing it at launch day, they'll reserve that little dessert for later.

constructorx 01-11-2019 01:24 AM

Quote:

Originally Posted by elguero (Post 27800280)
I also think the processor we so, was a Ryzen 5, but Ryzen 7 probably will be a 12 core, I don´t see AMD releasing 16 core just yet, maybe later on the year.

Don't get me wrong, AMD could/can/will release a 2 chiplet 16 core any time they want, but I just don´t see them doing it at launch day, they'll reserve that little dessert for later.

They may release a 12 core. They could also release more than one 16 core. That is the fun in speculating. We shall see soon enough.

bmgjet 01-11-2019 01:40 AM

Was expecting to see a bit more information.
But what we did see was promising.

Which was better efficency which was expected with 7nm.
Better multi-thread performance at lower clock speed.
Clock speed boost.
A time frame to wait so now if people are thinking about upgrading to a 9900k they can make the judgement call if its worth waiting 6 months.
And most of important, no hidden chiller.


There are still some big questions left un-answered like which model they demoed, And how much further can it clock. Im hoping they are showing the chip for its efficency and there is some good head room left instead of showing a chip that pulls same power with 200 point higher score.


Quote:

Originally Posted by elguero (Post 27800280)
I also think the processor we so, was a Ryzen 5, but Ryzen 7 probably will be a 12 core, I don´t see AMD releasing 16 core just yet, maybe later on the year.

Don't get me wrong, AMD could/can/will release a 2 chiplet 16 core any time they want, but I just don´t see them doing it at launch day, they'll reserve that little dessert for later.


They will hold it back to bin chips for it.

elmor 01-11-2019 02:08 AM

- Early ES
- Zen 2 is supposed to have IPC increases around 10-15% (https://wccftech.com/amd-zen-2-7nm-c...ncrease-rumor/)
- Peak power consumption was only ~75W (https://www.anandtech.com/show/13829...e-4-eight-core)

If Zen+ reaches the same score at 4.35 GHz and we assume a 10-15% IPC increase based off rumors, the chip would have been at around 3.78-3.95 GHz all core turbo.

Zen+ uses ~90W at 3.8 GHz (https://forums.anandtech.com/threads...#post-39391302) and ~110W at 3.9 GHz. That would mean 17-32% lower power at the same frequency from going to 7nm.

jclafi 01-11-2019 02:35 AM

I'm going w/ Zen2+.... My current RIG is really incredible, can handle everything i do so easy, so i can wait a year or so...

But when AMD release Zen2+ in 2020, i believe still in AM4 socket, i upgrade my R5 2600.

Good job AMD !

Storx 01-11-2019 04:11 AM

Interesting points over on Reddit, not being an experienced person on cinebench... i didn't know how it works with turbo boost, many mentioned that if you bench a 9900k stock it will run at 4.7Ghz all cores, so that leads me to believe that zen 2 is running 4.5-4.7Ghz, because SMT is about 10% better than HT, but the part that kinda concerns me is AMD would have brought the most binned chip they could find to the show, but one point that they keep bringing up is Lisa mentions Zen 2 competes core for core running the same with same hardware.... which is making many think this chip was running 4.7Ghz all cores to match the statement she mentioned.

Also this was leaked on AMD page.... "Łukasz Lis, The Demo Zen 2 CPU Is A Stepping B, 5D0108BBM8SH2_37(later revision, Quality Sample) Ryzen 5, 8core/16 threads, 65W TDP, at 3.7base clock , AM4 socket, hitting 4.5ghz"
http://ir.amd.com/news-releases/new....gh-performance

Dogzilla07 01-11-2019 06:24 AM

Quote:

Originally Posted by elmor (Post 27800340)
- Early ES
- Zen 2 is supposed to have IPC increases around 10-15% (https://wccftech.com/amd-zen-2-7nm-c...ncrease-rumor/)
- Peak power consumption was only ~75W (https://www.anandtech.com/show/13829...e-4-eight-core)

If Zen+ reaches the same score at 4.35 GHz and we assume a 10-15% IPC increase based off rumors, the chip would have been at around 3.78-3.95 GHz all core turbo.

Zen+ uses ~90W at 3.8 GHz (https://forums.anandtech.com/threads...#post-39391302) and ~110W at 3.9 GHz. That would mean 17-32% lower power at the same frequency from going to 7nm.

Yup the various pieces of information from different sources and the numbers are starting to come together.

If you take into account pre-brief score of 2023 compared to onstage score of 2057 from Ian's Anandtech article, and add the AMD page leak data to that from @Storx post above, what we end up getting is a 3.7GHz base, 4.5GHz single core boost chip that scored 2023 in CB 15 at the pre-brief, which in turn gets PBO oc-ed a bit and that's where the 4.6GHz single core boost comes from and the 2057 CB 15 number on stage.

IPC increase seems on the lower end, could that correspond to ram running on 2666MHz ? or increased global RAM latency due to the state of ES samples and chiplet design ?

Any additional news or info on the rumors that the infinity fabric is not gonna be tied to RAM speed this time around and is gonna have an uncore clock ?

white owl 01-11-2019 07:24 AM

Quote:

Originally Posted by Storx (Post 27800466)
Interesting points over on Reddit, not being an experienced person on cinebench... i didn't know how it works with turbo boost, many mentioned that if you bench a 9900k stock it will run at 4.7Ghz all cores, so that leads me to believe that zen 2 is running 4.5-4.7Ghz, because SMT is about 10% better than HT, but the part that kinda concerns me is AMD would have brought the most binned chip they could find to the show, but one point that they keep bringing up is Lisa mentions Zen 2 competes core for core running the same with same hardware.... which is making many think this chip was running 4.7Ghz all cores to match the statement she mentioned.

Also this was leaked on AMD page.... "Łukasz Lis, The Demo Zen 2 CPU Is A Stepping B, 5D0108BBM8SH2_37(later revision, Quality Sample) Ryzen 5, 8core/16 threads, 65W TDP, at 3.7base clock , AM4 socket, hitting 4.5ghz"
http://ir.amd.com/news-releases/new....gh-performance

You wouldn't want a binned chip if you're showcasing a future line up as the sample provided would be better than the rest. She also stated that it was only an ES chip and that's not the final speed. ES chips are always less than or equal to the final product.
A STOCK 9900k will NOT hit 4.7ghz all core, only 4-4.1Ghz all core with some spikes at 4.7Ghz. One you remove the TDP limit you get your 4.7Ghz which is an overclock even if it's enabled from the factory by the motherborad (just like MCE).

Zen2 beat the 9900k at 4.6Ghz vs 4.7Ghz and AMD actually allowed the INTEL system to be overclocked since Intel's TDP is 95w. Allowing it to use the 120w+ to hit 4.7Ghz all core says a lot.

[email protected]/4.6Ghz+/[email protected]/unlimited TDP

miklkit 01-11-2019 10:04 AM

Methinks in 7-8 months I'm going to be pretty excited. The biggest concern is memory prices. I want to buy more and better ram but not for $500.


Overall, this cpu is a big enough jump over the 1700 to make it well worth it.

Roaches 01-11-2019 01:49 PM

To be honest from a neutral standpoint the keynote was normal but was met with high expectation from the community due to people believing leaks from popular media channels which generated alot of hype unfortunately

I mean if you compare this to their last year Ryzen 2 release it was received as very normal as in not impressive or underwhelming but just positively a step forward as you would expect from incremental improvements.

Having seen the exposed package of Ryzen 3 does hint much potential positives and I can see why people are so hyped over it though really its just a normal day for the industry to showcase their latest developments.

My thoughts on Vega VII is rather mixed seeing that its a compute focused card being sold to gamers when I felt it would've been more positively received by investors if it was a Radeon Pro card aiming to undercut the Titan V in which It would be fine to price it high as a pro card with pro drivers for less than the competition offering. I'm personally in the mid range market so yeah it doesn't bother me.

Its just a normal day in the industry though with positive hints moving forward.

et7878789 01-11-2019 03:19 PM

i think ES CPU is running at 4GHz.
https://i.imgur.com/a63YnPO.png
https://i.imgur.com/gIw6Kph.png

Storx 01-11-2019 03:40 PM

Some futher info has been released on the AMD CES cinebench, someone took the footage of the cinebench run and matched it up with the power draw numbers and noticed that they dont line up at all, in the CES demo the people clicking the mouse actually are clicking to start the benchmark differently than what the screen displayed, in the video person clicking for the AMD machine actually is slightly before the intel person clicking, but in the video of the benchmark the start of the two benchmarks they start at the exact same time, but what is interesting is the wattage displays dont match the benchmark runs. If you open cinebench and start the benchmark you will see that the load will spike, then go to idle, then peg out for the test. In the video you see Intel dropping to a idle of 46.3watts during the middle of the run on the screen, and ryzen dropping to 67watts. So this shows that the people on the stage actually were not running the demo live...

The interesting part is the idle watts, ryzen seems very high for what it is 7nm, but new news told that the ryzen cpu is still engineer stages, that they are still perfecting the XMR boosting of the chip, so the machine was manually set all core to the "same" clock speeds of the intel counterpart. The actual clockspeeds were not released just the fact that it was manually clocked, so the chip was idling high due to this. There has been more than one source stating that it was the Ryzen 5 cpu with base of 3.7, but different sources say 4.5 vs 4.6Ghz, which sounds like they are still a long ways away from release on the chip, if they are still perfecting turbo boost on the cpu.

speed_demon 01-11-2019 03:51 PM

Ooh we got ourselves a good old fashioned conspiracy here! ;)

gapottberg 01-12-2019 09:41 AM

As others have mentioned...my judgement is also tied to what exactly we are looking at. If AMD just showed a $200 Ryzen 5 edging out the $500 9900k as some have speculated then this is a huge victory for AMD. That was a head to head with same number of cores and likely slower clocks at possibly half the TDP as it's speculated that chip may be a 65watt chip. That means better IPC and better gaming capability or at the least similar.

Clock speed isn't King in gaming. It's work per unit time, which is a function of both clock speed and IPC. Clock speed alone is about as useful as having RPMs for a car engine and trying to guess what speed it's travling at. You need more relevant info to make any claims about effectiveness in games. What this demo shows is that since core counts and thread counts are the same... AMD has officially caught up to stock Intel in performance core for core...regardless of what clock speed actually is. There may be head room to surpass it...and there may not be, but regardless of that they showed here it's possible to match them. Something they could not do a year ago. They needed more cores and thread to beat Intel in multi core tasks. Now they don't.

Assuming the above speculation is correct, that would suggest the R7 and R9 class chips will indeed have a second chiplet and will likely be 12 and 16 core CPUs. They would blow anything else out of the water in terms of consumer grade production and content creation for casual users and streamers as they have double the cores and equal performance to Intel core for core.

JackCY 01-12-2019 10:20 AM

They are pushing the CPU industry forward, unlike Intel who sits and cashes in. So opinion is positive as long as the products remain a better value than competitor.

Gilles3000 01-12-2019 10:48 AM

Quote:

Originally Posted by miklkit (Post 27801038)
Methinks in 7-8 months I'm going to be pretty excited. The biggest concern is memory prices. I want to buy more and better ram but not for $500.

The prices don't seem to be rising anymore, and the regular 2400mhz-3200mhz stuff seems to be dropping in price steadily.

Still unfortunate that the B-die stuff isn't dropping back to 2016 pricing tho...

Undervolter 02-05-2019 02:42 PM

Quote:

Originally Posted by jclafi (Post 27800362)
I'm going w/ Zen2+.... My current RIG is really incredible, can handle everything i do so easy, so i can wait a year or so...

But when AMD release Zen2+ in 2020, i believe still in AM4 socket, i upgrade my R5 2600.

Good job AMD !

I just ordered 2600 and will do the same, but when AM goes EOL.

Quote:

Originally Posted by miklkit (Post 27801038)
Methinks in 7-8 months I'm going to be pretty excited. The biggest concern is memory prices. I want to buy more and better ram but not for $500.


Overall, this cpu is a big enough jump over the 1700 to make it well worth it.


I read an article, that RAM prices are expected to drop >10% in 2019, which is why RAM manufacturers are cutting production yields, in an attempt to prevent big price drops. At any case, it is ridiculous, that 3200 low latency RAM costs more than the CPU... It's a ripoff, plain and simple.

Good to see you both again, BTW.

----------------------------------------------------

About CES, while it is good to see AMD attacking even more Intel, i can't help but wonder, where this thing with cores will go. In the past years, i read Amdahl's law and i can't help wondering, how is a consumer going to benefit from 12 or 16 cores, unless he is able to run one of the very few, perfectly parallelized and divided in subworks tasks, like video encoding. The industry needs a revolution in IPC, not so much more cores, IMHO. 8c/16T aren't going to be fully used in games before "IPC becomes obsolete" for the forseable future. By the time this happens, the IPC of a Ryzen 1st, 2nd, 3rd gen will be "ancient". Sure, they can make 6 and later 8 cores, entry level, but how many people will actually use them to their potential? And you can "thank" Intel for that, since for 10 years, they 've sold like gold 2 and 4 core CPUs and game producers, even if they did manage to parallelize perfectly for 12,16,32 threads, will simply not do it, because so many potential customers, still run 4,6 and 8 threads.

miklkit 02-06-2019 09:49 AM

1 Attachment(s)
Good to see you back!



Multithreading is happening more and more. I gave up on my FX when I started seeing gaming loads in the 80-100% range. Ny Ryzen 1700 is holding up well so far but more performance will be needed in the future. This is what a Unity engine based game looked like on my rig 3 days ago. A quad would be saturated.

Undervolter 02-07-2019 08:20 AM

Quote:

Originally Posted by miklkit (Post 27841230)
Good to see you back!



Multithreading is happening more and more. I gave up on my FX when I started seeing gaming loads in the 80-100% range. Ny Ryzen 1700 is holding up well so far but more performance will be needed in the future. This is what a Unity engine based game looked like on my rig 3 days ago. A quad would be saturated.

Ah, things have evolved. 16 threads and even load across all of them! Who would have thought... I guess when Intel moves to more cores, everything starts moving towards there too. But, 6 cores aren''t going to be saturated any time soon. By the time you really need 8, you will be running DDR5 and whatever comes after Zen 3... I 'd buy 8 core Ryzen if i intended to stay with it for 5 or more years. Which i probably should, only now i know i would be kidding myself. I have a 2600 arriving next week and i will probably grab a 65W 8 core in 2020, of whatever Ryzen 4000 will be then. At that point, i should stick with it for many years (and Win7). I was tempted of ordering 1700 now, but the low stock clock together with the lack of will to overclock and the unknown RAM behaviour of 1st gen, made me go for the 2600.

I hope i see a general single thread improvement on desktop operations and some programs, as it's the main excuse i can find to upgrade from the [email protected] I still play the same few games i used to play 2 years ago (mainly Skyrim and Europa Universalis) and had no real need to change anything. I actually found 2 videos of Ryzen with Skyrim and one seems the same FPS i get after the tweaks i apply (now i am convinced it's a limitation of the game engine that simply drops FPS on specific spots), the other is actually slower than the FX (and imagine i run still with GTX750Ti), because he doesn't apply tweaks. I will probably earn something in Europa Universalis IV, which is very CPU intensive... I found a youtube video and i think it runs a bit faster... Anyway, i try to find various excuses to rationalize my impulsive order, but truth be told, i should have stayed with FX at least until Zen 2 or better yet, until Ryzen 4000. But, upgraditis is a disease, isn't it... :p Oh, yeah, i played a more recent game called "Shadow tactics", but the FX was something like 45% loaded across cores. Buttery smooth.

miklkit 02-08-2019 12:54 PM

All CPUs have microcode built into them and I have found that the old instruction sets in FX is the main thing holding them back. FX can not run the latest SSE. For this reason I
have decided to update the cpu every 2-3 years just to get the lates instruction sets.


When I first got this Ryzen 1700 up and running at speed I compared the [email protected] to the [email protected] in old and new games. The FX was better in heavily modded DX7 and DX8 games. The FX and Ryzen were more or less even in DX9 games, but Ryzen was 20fps better in DX11 games. That's the microcode difference in action.


I will be getting a 3xxxX cpu next fall and have decided to get an X570 motherboard as well. Unfortunately this Biostar X370 GT7 has spoiled me. I might have to get a GT9 to get all the features I have gotten used to.

Undervolter 02-09-2019 01:42 AM

Quote:

Originally Posted by miklkit (Post 27844878)
All CPUs have microcode built into them and I have found that the old instruction sets in FX is the main thing holding them back. FX can not run the latest SSE. For this reason I
have decided to update the cpu every 2-3 years just to get the lates instruction sets.


When I first got this Ryzen 1700 up and running at speed I compared the [email protected] to the [email protected] in old and new games. The FX was better in heavily modded DX7 and DX8 games. The FX and Ryzen were more or less even in DX9 games, but Ryzen was 20fps better in DX11 games. That's the microcode difference in action.


I will be getting a 3xxxX cpu next fall and have decided to get an X570 motherboard as well. Unfortunately this Biostar X370 GT7 has spoiled me. I might have to get a GT9 to get all the features I have gotten used to.

I see... Since i don't intend to move to Win10 any time soon and since i don't play much new games, i don't have the same eagerness for constant upgrades, but, i hope i will at least feel like having a faster desktop and program launching/execution. I use often some programs that are saturated in single thread with the FX. So, i hope that i will put the Ryzen in and say "wow, that was so faster"!. This will make me happy enough, because, otherwise, the only game that i will probably want and would benefit a lot from Ryzen isn't out yet. It's Elders Scrolls VI, which is still "under construction" and if it's similar to Skyrim, i will probably get it.

miklkit 02-09-2019 11:02 AM

You won't notice anything on the desktop. Ryzen might even be slower.



The single thread performance of my FX was about the same as the 2600k while the single thread performance of my 1700 is about the same as a 4790k. At stock this 1700 ran at 3 ghz and boosted to 3.742 ghz but after OCing it performs better than less than 200 mhz would indicate. And volts aren't that bad. This one hits 1.394v in light loads and 1.356v in heavy loads. The LLC is set to "auto" and I'm fine with that.

Undervolter 02-10-2019 12:35 AM

Quote:

Originally Posted by miklkit (Post 27846132)
You won't notice anything on the desktop. Ryzen might even be slower.



The single thread performance of my FX was about the same as the 2600k while the single thread performance of my 1700 is about the same as a 4790k. At stock this 1700 ran at 3 ghz and boosted to 3.742 ghz but after OCing it performs better than less than 200 mhz would indicate. And volts aren't that bad. This one hits 1.394v in light loads and 1.356v in heavy loads. The LLC is set to "auto" and I'm fine with that.

What! Slower??? Great... Maybe i made a hole in the water at the end... I intend to leave anything to auto too, which is also one of the reasons why i preferred the 2600 over the 1700. I don't do much video encoding anymore, so the extra 2 cores would be wasted. I will keep FX rig for video encoding, if needed. From what i saw, the 2600 is on par with 1600X, which is 3.6Ghz (both at stock). So i thought "better a cpu at 3.6Ghz at stock, than at 3Ghz with 2 more cores). I will be very disappointed if it turns out that it feels slower than Vishera.

miklkit 02-10-2019 05:54 AM

FX is fast on the desktop. My 1700 was quite slow at first but after some tweaking it got better. Latency is a problem and Ryzen can be finicky about it while FX doesn't sweat it and just works.



I thought I had a good running FX system until I pulled the hard drives and put them in th Ryzen rig. Nothing but problems! But once the driver conflicts and latency issues were taken care of it got up to speed. Still not sure if it is as fast as FX.

Undervolter 02-10-2019 06:55 AM

Quote:

Originally Posted by miklkit (Post 27847174)
FX is fast on the desktop. My 1700 was quite slow at first but after some tweaking it got better. Latency is a problem and Ryzen can be finicky about it while FX doesn't sweat it and just works.



I thought I had a good running FX system until I pulled the hard drives and put them in th Ryzen rig. Nothing but problems! But once the driver conflicts and latency issues were taken care of it got up to speed. Still not sure if it is as fast as FX.

I ll be damned! I hate sluggish feeling on the desktop. Actually, i read about 1st gen Ryzen latency issues and that's another reason why i went with Ryzen2600, hoping that this would be less of an issue. I am crossing my fingers. I would hate it if i made a Ryzen build, only to find myself relegating it to secondary machine for gaming only, while i keep the FX for general purpose rig. Duh.

The Pook 02-10-2019 07:49 AM

Quote:

Originally Posted by miklkit (Post 27847174)
I thought I had a good running FX system until I pulled the hard drives and put them in th Ryzen rig. Nothing but problems! But once the driver conflicts and latency issues were taken care of it got up to speed. Still not sure if it is as fast as FX.


https://i.kym-cdn.com/entries/icons/.../729/large.jpg

jclafi 02-11-2019 06:14 AM

I have 67ns memory latency w/ my 3200 RAM.

My R5 2600 is WAY FASTER on Desktop and Gaming compared to my old FX 8350. Once overclocked the performance is even better! I still have my FX system w/ SSD and one GT730 DDR5 card. My wife uses the system and love it !

No sense talk !

miklkit 02-11-2019 09:09 AM

Quote:

Originally Posted by The Pook (Post 27847252)


Yeah, the driver conflicts were between Macroshaft, Creative, and AMD. The FX ignored them but they had the mouse lagging badly and the sound cutting out with the Ryzen 1700. It took me a month to get it all sorted out.

Undervolter 02-11-2019 05:06 PM

Ok, now that i 've installed everything, i can say that FX is really close in general feeling, but the Ryzen feels a bit snappier on the desktop. And it might be due to better SSD performance. Opening the menus in the control panel aren't faster. They feel just like FX. In general, i find a small improvement, but nothing earth shattering. Some applications seem to launch faster too.


All times are GMT -7. The time now is 01:26 PM.

Powered by vBulletin® Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.

User Alert System provided by Advanced User Tagging (Pro) - vBulletin Mods & Addons Copyright © 2020 DragonByte Technologies Ltd.
vBulletin Security provided by vBSecurity (Pro) - vBulletin Mods & Addons Copyright © 2020 DragonByte Technologies Ltd.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2020 DragonByte Technologies Ltd.