well Tweaktown guys are saying x1800 Ryzen in gaming will performance like a 6700k, but applications that can utilize more core it will go against a 6900. Don't know how true this is, as I was expecting x1800 to go up against 6 core Intel Broadwell 6 core cpu's.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Leaked AMD Ryzen Benchmarks?
- Thread starter HardOCP News
- Start date
OrangeKhrush
[H]ard|Gawd
- Joined
- Dec 15, 2016
- Messages
- 1,673
And I get flamed for pushing Haswell performance ie:
1800
1700 = 5960X
1600
1500 = Octocores
1400 = 4770/4790
1300 = i5's
1200
1100 i3/15/Pentiums
all that is rather close to KBL anyways
From the very vague blanket answers to blanket questions the sample shows performance mostly up from the 5960X, in some instances it matches or beats a 6900K but in general 5% or so slower. If you offered me that 5 years ago I would have snatched it from your hands and run.
This video highlights the state of problems that Intel is facing, kind of problems that flow from bad management, if there is a company that can show the effect of bad management it is AMD, under Hector Ruiz AMD went from 40 to the share value down to flatline in 4 years on the basis of cheap operations, selling assets, employing non engineers, bad acquisitions.
That said AMD over the last 3 years under Su in particular has focused on only employing the best available which not only includes Jim Keller but engineers from Intel, Nvidia, IBM, ARM, AMD under better leadership have shown massive improvement. Do I think that Ryzen is the scourge of the fanboy, yes it will cause many tears to fall, but have no sympathy for that. Said it many times Intel is going nowhere right now, laid off 12000 with more to follow soon, may end up 5 generations on 14nm and by 2020 Intel will be solely focused on data center parts, this is not encouraging at all. Neither is that there is no talk of lowering prices as Barrons reported Intel's R&D is running serious debt, Intel dropped billions (not 1 or 2 but more like 30bn) on bad investments and shareholders will never devalue their shares to get into a price ware with a company that is actually not the main threat anymore. ARM/Qualcomm, Nvidia are starting to break into Intels free money domains, aMD is also going to try its luck in HPC, data centers to. It is not a easy time for Intel corp, and 14 years of having it easy, now they will have a test. What isn't helping is if your best engineers are wanting to abondon ship
If AMD give me an option to buy a low cost octo core enthusiast grade CPU and Intel cannot then I will take
x1800 from what I have been reading is doing pretty good against 6900. If its performing like 6700k in games that is actually a plus since most games don't do well above 4 cores.
OrangeKhrush
[H]ard|Gawd
- Joined
- Dec 15, 2016
- Messages
- 1,673
If it does then AMD have been doing some serious sandbagging
You got flamed from saying IPC was at Haswell level lol, IPC is not performance two different things.
Bad managment from an engineering perspective after getting its handed to them for 5 years in the tech world is why AMD is in the position its at, not because of Hector Ruiz, it might have started with him but it got much worse later on.
And the rest of that video, yeah its a pipe dream, Intel has much more capabilities than AMD, for now it looks like they just have to drop pricees, and this explains why Intel is expecting less margins in the coming future.
Intel's R&D debt, who ever wrote that article doesn't know shit about making chips, cause R&D is a fixed cost and shouldn't be factored into as a marginalized item and the investements Intel aren't bad, just that investors won't see the fruits of those investments for a few years something that journalist doesn't know about because he doesn't understand how integration of large tech companies happen, its a painful process. Just ask AMD and ATi..... What its been more than 10 years and they are still trying to figure things out.
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,678
CPC about Ryzen gaming performance.
Someone above said "AMD is also going to try its luck in HPC", not so fast. Zen Architecture Good for Desktop and Server Market But Not HPC
OrangeKhrush
[H]ard|Gawd
- Joined
- Dec 15, 2016
- Messages
- 1,673
If performance is the product of IPC in the application in question then performance derives from that. All things considered Haswell type performance accross the board is a plausible outcome.
From what they have shown thus far yeah from Carand's early benchmarks of and ES, and recent leaks, looks to be a bit less than Haswell in IPC.
IPC is part of performance but not all.
There are tests where Bulldozer can match Haswell in multi threaded applications, end result its performance in those tests are damn good, but its IPC still sucks......
Its everything I and others here have been stating, you took IPC, at least what you thought was IPC from AMD's showings, which it wasn't IPC, then extropolated it based on IPC gains AMD is expecting, and then posted on it. Which is just incorrect. And you are still trying to say that lol.
The whole point of what amd doing here is driving customer up to their 6 core and 8 core chips. There is no reason in hell your 4 core should be clocked less than your 8 core. That is a pure marketing decision. I am sure the 4 core chips will probably overclock in the 4ghz range but AMD has more profit sitting in the upper end chips, and you see higher clocks there.
Note x1400 does have XFR, who knows with nice cooling it will be staying in the 4ghz range. We will only find out when the reviews come out.
drescherjm
[H]F Junkie
- Joined
- Nov 19, 2008
- Messages
- 14,941
Agreed, This part does not make sense. Unless they cherry picked the best cores and used them in the 8C. Does the 8C use 2 x 4C dies or is the 4C an 8C with 1/2 the cores disabled??
Last edited:
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,678
There is only one Zen die. Its a ~200mm2 with 2 GMI links, dual channel and 2 CCX clusters with 4 cores in each and their own non shared 8MB L3.
ecmaster76
[H]ard|Gawd
- Joined
- Feb 6, 2007
- Messages
- 1,150
Odds are the reduced core parts are from older steppings and/or failed validation on all 8 cores. The lower clock speed targets are to let them sell as many parts as they can
We'll probably see higher binned 4c parts in a few months.
But since they are all unlocked AMD is just making us choose between two brackets:
-more expensive, but more cores and guaranteed clock speed
-less expensive, but less cores, and silicon lottery for clock speed
If the cut down parts can have cores unlocked and/or OC very well they will be incredibly popular and will spawn a few dozen threads on this subforum
There were rumours awhile back that for some reason AMD was having problems with the 4C.
Cheers
lose in games? What does that actually mean? 1 fps less or 2 min fps? Please enlighten me I have a 6850k, I sure as hell feel no pain points with this processor compared to 7700k. Zen 8 core with 3.6ghz base really going to be a problem for games? With decent cooling it should be hitting 4ghz easily. Common now.
drescherjm
[H]F Junkie
- Joined
- Nov 19, 2008
- Messages
- 14,941
I am not sure how you could have problems with 4C but at the same time have 8C working fine.
Although the idea that a 4C is just an 8C with 1/2 the cores disabled can explain the lower clocks on 4C chips. I mean if the cores are that bad that 1/2 of them did not make the cut it is understandable that the frequency had to be lowered.
Last edited:
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
Hm?
Magic Hate Ball
Limp Gawd
- Joined
- Jul 28, 2016
- Messages
- 366
I think he's implying that it'll take watercooling to get a stock 1800X to get to its 4.0ghz boost clock...
Which I don't think I agree with.
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
Yeah, i believe the "big" cooler is simply Wraith.
pendragon1
Extremely [H]
- Joined
- Oct 7, 2000
- Messages
- 52,049
yup just looks like a redesigned shroud on the regular wraith cooler. not sure which thread its in but ppl are complaining that the cooler is capable of 150w and they saw that as bad "wtf does a 95w chip need a 150w cooler blah blah blah"....
That's what Tweektown guy in the video stated. 6:20 mark, 4.0-4.3 with water. Air cooling (even the wraith) is going to be less, and it looks like they confirmed that with their source. Ln2 to get higher than 4.3 up to 5.0 with LN2.
Pretty much what it sounds like is the process is going to limit the clock speed on Ryzen, Intel's 14nm is more mature and more advanced, thats probably where Intel is going to stay on top for now.
Gotta remember TDP for AMD and Intel are different, Intel is more real world power usage, AMD's rounds down on their measurements. But would be interesting if AMD goes to more of an Intel type measurement or stick with what they had before.
Last edited:
Yeah I am not sure as well, and having difficulty finding those old sources.
But this is a complex CPU and could come down to the various functions/caching at a 4C level if it was just based on the 8C.
In theory and if ideal you just cut the cache but it may have some implications, then there is the prefetch/'neural net' prediction (hate that name tbh)/SenseMI/SMT/etc and how this all ties together.
I think we need to see more details from AMD to get a feel why there is a discrepancy, but so far that rumour about the 4C seems to be right so far.
Cheers
understandable but here we are coming from 3.12 initial ES samples to 3.6 base and possible 4+ Ghz on Water? I think that is pretty decent for what they have come up with so far on Zen's initial launch.
pendragon1
Extremely [H]
- Joined
- Oct 7, 2000
- Messages
- 52,049
ths isn't new news. Su said it cleary in the presentation, better cooling = higher speeds.
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,678
https://videocardz.com/65913/how-fast-is-ryzen
The ST numbers looks really, really weird. Ryzen performs the same at 3.4, 3.7 and 4.0ghz for example, while quite slower at 3.7Ghz. Looks like someone ran both the 4 and 6 core at 4Ghz as well.
The ST numbers looks really, really weird. Ryzen performs the same at 3.4, 3.7 and 4.0ghz for example, while quite slower at 3.7Ghz. Looks like someone ran both the 4 and 6 core at 4Ghz as well.
Last edited:
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
If you have spent 10 seconds you would notice that 3.5Ghz 6900k is 11% slower than 3.5Ghz 6800k. It is an issue with 3DMark's test core scaling, if anything.
But yes, either the 4Ghz result is not actually at 4Ghz, or default turbo mechanism on six and eight core forces it to run at about 3.5Ghz. If it is the latter, then the throughput in this test closely matches Broadwell.
That is basically the description of first batch of Broadwell-Es, actually. I mean, if it is truth i can foresee folks being disappointed with 4Ghz bin only overclocking to 4.3Ghz.
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,678
The Intel numbers comes from god knows where. The MT numbers seems to come from Toms. But not the 7700K numbers and 4.2Ghz 6800K is VGZs own. There are no ST numbers on Toms.
I have a feeling the ST numbers is made up. MT divided with cores.
Broadwell E is still on the older process though...... Intel fin design changed with Skylake.
Agreed. There is a famous naysayer Juan who comes up with theory after theory to explain Ryzens excellent 8 core clockspeeds. He blames lower 4 core clock speeds on a flawed process, when in fact it is a marketing decision to sell their higher 6 and 8 core Ryzens. Later in about 3-4 months AMD will not disable cores to make a 4 core . They will have another higher clocked 4 core X product with significantly higher clock speeds than what is mentioned now.
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
That's exactly what they are, and that's why i did math with original numbers.
Really? I believe Broadwell and Skylake are on the same process, it is just that Kaby that messes it up.
Let's first see how it overclocks, okay?
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,678
VGZ now edited their images. Showing Intel parts with full ST boost clocks in an MT bench. This is getting more and more hilarious.
Why are you quoting Juanrga?? He is all over the place with his theories of the 14LPP process AMD used for Ryzen. He consistentlu has underestimated clock speeds on the 8 core Ryzen starting with a 3.0 GHZ estimated maximum base frequency. Now we have a 3.6 GHZ-4.0 turbo as flagship that under water can go at least as high as 4.3 GHZ on all 8 cores with a quality cooler. Now he blames lower clocked 4 cores on the 14LPP process that has produced higher frequency 8 cores. he is out of his bird. The real reason for the 4 core frequencies is that AMD initially wants to sell its higher margin 6 and 8 cores. In a few months after initial launch there will be 4 core X chips on Ryzen clocking higher than the 1800X. Right now 4 cores are all coming from disabling cores on 8 cores. AS the process becomes more efficient this will stop . More and More 4 core X chips will be produced on its own line. They will kick ass on Intel at a lower price.
Last edited:
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,678
Why are you not reading the twitter? Try read what CanardPC says. Its not that complex is it?
Check semiaccurate.com forum for cpu. See the interchange between DresdenBoy , Schroeder, and Drunkenmaster, all industry stalwarts who have a good knowledge of AMD. They demolish Juanrga's twisting of data and misrepresenting facts. as far as Canard PC goes. They have NOT done any testing on final silicon at least not in their public reviews. Their only 4 core was an A0 engineering sample that Juanrga states was not an engineering sample.
Spirit_Retro
Limp Gawd
- Joined
- Apr 1, 2010
- Messages
- 466
That thread is entertaining. In a train-wreck sort of way.
It amazes me... how folks can argue so hard for things that are so meaningless. I stay out of it because I'm still having fun. Maybe I got old.
After eye surgery tomorrow, they tell me I'll actually be able to see my screen again... Here's hoping.