Leaked AMD Ryzen Benchmarks?

razor1 said:
I don't really care about the clocks, I have already stated numerous times IPC might be on Intel's side, but SMT performance evens the scales for AMD, so end results Intel and AMD should be on equal footing for the most part. If Intel doesn't cut prices, Intel is going to to have a tough time convincing people to stay away from AMD for ALL their desktop i7, i5, i3 procs.
Click to expand...

intel has plenty of wiggle room, i don't think they'll really touch their current gen prices all that much. they'll probably wait a few months before they do anything concrete, in the meantime they'll continue to post teasers and rumors about stuff they're releasing down the road which will probably keep a good percentage of people willing to wait instead of buying now. either way their bread and butter is the i3/i5 sales and because AMD isn't releasing their mid/low end til later there's not much to worry about on that front. if the benchmarks reach AMD's goals of threatening intel you'll see intel reacting to it just before the 6/12 and 4/8 core chips hit the market.
 
That is true, but its only about 1 quarter of time lol, not too much. Skylake E has to be priced competitively and then we got Cannon Lake, OEM's really won't touch Ryzen outside of Boutique stores, till Ryzen + APU's come out. So yeah Intel has some time, but not too much.
 
razor1 said:
That is true, but its only about 1 quarter of time lol, not too much. Skylake E has to be priced competitively and then we got Cannon Lake, OEM's really won't touch Ryzen outside of Boutique stores, till Ryzen + APU's come out. So yeah Intel has some time, but not too much.
Click to expand...
I really think intel is just going to bury their heads in the sand and hope that marketing prevents Ryzen from taking off. i do not expect Intel to deviate at all from their current plan unles either market share drops or investors start jumping ship. Likely it would take both of those. Most likely we won't see to many OEM partners sprouting Ryzen this year at all as Intel probably locked in contracts already. This will be a wait and see game. People thinking intel will drop their prices just look at Skylake vs Kaby lake nearly identical processors Intel is selling both side by side with only a $10-$20 difference in prices. This is a very complacent and arrogant company that is likely to do nothing unless the market absolutely demands it.
 
Presbytier said:
I really think intel is just going to bury their heads in the sand and hope that marketing prevents Ryzen from taking off. i do not expect Intel to deviate at all from their current plan unles either market share drops or investors start jumping ship. Likely it would take both of those. Most likely we won't see to many OEM partners sprouting Ryzen this year at all as Intel probably locked in contracts already. This will be a wait and see game. People thinking intel will drop their prices just look at Skylake vs Kaby lake nearly identical processors Intel is selling both side by side with only a $10-$20 difference in prices. This is a very complacent and arrogant company that is likely to do nothing unless the market absolutely demands it.
Click to expand...


i mean you're spending 10-20 dollars for a better igp, thats pretty much it which yeah sure doesn't matter to the enthusiast but matters to the mobile/oem market. i'm not a fan of intel's marketing but i wouldn't really call it arrogant, i'd call it they fucked up and put themselves in a hole they can dig themselves out with their guaranteed release schedule. kaby lake ends up being one of those, "oh shit we didn't predict that, we need a new IGP for the mobile market" processors. either way ryzen should destroy the market for past gen cpu's. why it still costs 200+ dollars to buy a 2500k or 2600k makes absolutely no sense.. i just hope i don't end up having to wait til may or june for the 1400x..
 
razor1 said:
I don't really care about the clocks, I have already stated numerous times IPC might be on Intel's side, but SMT performance evens the scales for AMD, so end results Intel and AMD should be on equal footing for the most part (win some loss some depending on the application and what the application needs are based on IPC or multi-threaded performance). If Intel doesn't cut prices, Intel is going to to have a tough time convincing people to stay away from AMD for ALL their desktop i7, i5, i3 procs.

They probably tested it before hand, I'm sure they just didn't expect it to happen or expect it to crash, ya don't go in a promotional setting for new hardware, without testing that hardware first. So its not about the crash per say, its about are most samples going to be stable above 4.0 ghz without water.
Click to expand...


Yea I saw the video. but in the video they applied the same setting that was on one core to all 8 cores of 4.1ghz without changing anything else. I was scratching my head, no wonder it failed and probably wasn't stable. I think they probably had a quick run before and didn't fully check for stability. Has happened to me a few times when I am overclocking. It would run the same bench 4 times and I think its stable and then 5th try it will crash until I give it another step up in voltage.

That is what I think happened here, Pros that they were so cool with some awesome software hahahaha. Ran it once before live demo and then tried the same thing again, ooops..
 
Presbytier said:
I really think intel is just going to bury their heads in the sand and hope that marketing prevents Ryzen from taking off. i do not expect Intel to deviate at all from their current plan unles either market share drops or investors start jumping ship. Likely it would take both of those. Most likely we won't see to many OEM partners sprouting Ryzen this year at all as Intel probably locked in contracts already. This will be a wait and see game. People thinking intel will drop their prices just look at Skylake vs Kaby lake nearly identical processors Intel is selling both side by side with only a $10-$20 difference in prices. This is a very complacent and arrogant company that is likely to do nothing unless the market absolutely demands it.
Click to expand...

Well you see, Ryzen is already selling like like hot cakes. This without even a review. I don't think intel marketing is going to do shit..
 
NKD said:
Well you see, Ryzen is already selling like like hot cakes. This without even a review. I don't think intel marketing is going to do shit..
Click to expand...
We literally have no idea how many units have sold through preorders. AMD has not released that number. Preorders could have been a small number to boost the appearance of high demand. Not saying that is what happened but we don't know at all.
 
NKD said:
Wait you squashed it that it wont matter, but how is something that doesn't matter so painful? We are talking about electronics here not a broken relationship that sure doesn't matter but still hurts lol.
Click to expand...
Because my workload is actually memory latency sensitive.
GroundZero said:
You are mistaking timings with total DDR4 latency. You just don't get it. Am I going to have to go over it all again here because you forget yesterday when you fall asleep?
Click to expand...
Exactly, and i have shown you what Sandra reports as total memory latency for Intel CPUs. Are you going to address elephant in the room? Right you won't.
This are actually bad timings, but 1T command rate explains why bandwidth efficiency is so high.
NKD said:
I could counter your argument by showing you MSI game boost dial picture that sets 1800x to 4.4ghz if you put the dial to number 11.
Click to expand...
You would actually prove his point in the process, the 11 number is unrealistic on Broadwell-E.
os2wiz said:
In single threaded apps with XFR enabled Ryzen can overclock a single core to the point it can outperform a Kaby Lake 7700 in single thread. I saw the stats over on that forum you no longer post on.
Click to expand...
Yeah, in a single workload that fits entirely in Zen's L2 cache but does not fit into 7700's (that by th
Presbytier said:
Actually if all the leaks are to be believed it may not actually be behind in IPC but equal.
Click to expand...
To Broadwell, it may. To Skylake, nah.

With that said, of course i am going to be critical of every rumor, because at this point anything past: "Broadwell IPC, worse than Broadwell-E clocks, way better price" is overhype.
 
Presbytier said:
We literally have no idea how many units have sold through preorders. AMD has not released that number. Preorders could have been a small number to boost the appearance of high demand. Not saying that is what happened but we don't know at all.
Click to expand...

I doubt it. This was not the case with polaris, or anything. This isn't just hype. People are actually buying this. AMD is reported to have 1 million units around the globe in stock on march 2nd. They must have confidence they are going to sell shit load or they wouldn't be making it so many.
 
Presbytier said:
I hope it is up tomorrow, but there has been lots of confusion if it is up on the 28th or march 2nd. I can't seem to find any reviewer saying for fact or AMD.
Click to expand...

from what I understand reviews on the 28th, release March 2nd.
 
Magic Hate Ball said:
This does seem odd that he would state such specific numbers as a person who received an official review sample. That sounds very much like an embargo breach.
Click to expand...

So true. Unless he is absolutely lying, this is the last time he gets a fat review kit from AMD. LOL, May be he is pissed his sample sucks.. bahahahha
 
Magic Hate Ball said:
This does seem odd that he would state such specific numbers as a person who received an official review sample. That sounds very much like an embargo breach.
Click to expand...
None of the numbers here reference review sample, however. It is an embargo breach if he does not lie, though. OTOH if the TDP thing is true, i will be pissed off.
 
lolfail9001 said:
None of the numbers here reference review sample, however. It is an embargo breach if he does not lie, though. OTOH if the TDP thing is true, i will be pissed off.
Click to expand...

To me he clearly is talking shit about AMD here. When you have a reviewer going out and clearly stating everything negative and in the way he is saying it with a negative tone, you know he doesn't give two cents about embargo, lol.

On the other hand we know AMD measures TDP different than Intel. But it seems like Ryzen power consumption is still below intel at similar clock rate.
 
I don't think you will see more than 3.8/3/9 All core overclocks on commercial coolers, I think you may be lucking with 4Ghz or slightly higher with a high end custom loop and that will cost a lot. Under phase change apparently to get 4.6ghz stable you need a lot of voltage, the LN2 was over 1.8V for 5.2Ghz so yeah, there is no way this pushes 24/7 north of 4Ghz on any commercial grade cooling.
 
lolfail9001 said:
None of the numbers here reference review sample, however. It is an embargo breach if he does not lie, though. OTOH if the TDP thing is true, i will be pissed off.
Click to expand...

Stating something about "other reviewers (like us) mostly also don't manage that level of OC" in regards to specific clock speed is very much against an embargo IMHO.
 
NKD said:
On the other hand we know AMD measures TDP different than Intel. But it seems like Ryzen power consumption is still below intel at similar clock rate.
Click to expand...
I will wait for actual measurements, if you will. That chinese leak does not have sensible numbers in that regard. In Prime95 sort of loads Ryzen should consume less. I am interested about other loads.
Magic Hate Ball said:
Stating something about "other reviewers (like us) mostly also don't manage that level of OC" in regards to specific clock speed is very much against an embargo IMHO.
Click to expand...
It may just be, but then again, Volker did sound pissed off.
OrangeKhrush said:
I don't think you will see more than 3.8/3/9 All core overclocks on commercial coolers, I think you may be lucking with 4Ghz or slightly higher with a high end custom loop and that will cost a lot.
Click to expand...
Phenom III indeed
 
Tzeh Pesh said:
To continue being cynical / contrarian / a bastard... Positive leaks, "MOAR! GIMME!" Negative leaks, "OMG! EMBARGO!"

NDA can't lift soon enough. Can't remember the last hardware release that had me this eager to read the reviews.
Click to expand...

Same here. I'm not ready to upgrade but I'll be damned if I'm not excited about all this. My last AMD system was the XP 2500+ Barton and that was one of my favorite systems I've ever had.
 
Looking around the internet, the defacto fall back has been wait for Skylake X, wait for cannonlake expecting really much different.

Skylake X will be more very expensive CPU's with Skylake generational IPC gains but low clockspeeds and more cores, for a very small niche market, if you think Intel will give you a $320-499 octo core, you must be tripping. i7 7740K on the new X platform is really tripping balls.

Cannonlake, yay smaller and smaller percentage die shrinks with less gains, it is almost like the performance outside some tweaking is limited by the "core uarch" itself. Cannonlake may end up 5% faster than SKL/KBL which is hardly exciting, base clocks will go up to FX 9550 stupidity levels and we will still have need to spend on high end boards just to keep it from combusting. or go non K and get dramatically less clockspeed and notably less performance.

These are the kind of moves to limit loss without cutting your profit turnovers in half and hoping that people buy into this sense of alliegence crap like it is a patriotic duty.
 
NKD said:
Well you see, Ryzen is already selling like like hot cakes. This without even a review. I don't think intel marketing is going to do shit..
Click to expand...
I think this go round Intel talking shit about AMD and trying to strong arm reviewers will backfire.
 
OrangeKhrush said:
Skylake X will be more very expensive CPU's with Skylake generational IPC gains but low clockspeeds and more cores
Click to expand...
Rumor is on the street that Skylake-X 8 core already beat 6950X record in Cinebench. So, it may be very expensive, but it sure is fast.
OrangeKhrush said:
if you think Intel will give you a $320-499 octo core, you must be tripping.
Click to expand...
If Intel thinks it can get away with $1k octo core again, IT must be tripping.
 
LOL all these leaks but no leaks about the actual review dates haha. I'll be pissed if this madness lasts for 4 more days.
 
NKD said:
LOL all these leaks but no leaks about the actual review dates haha. I'll be pissed if this madness lasts for 4 more days.
Click to expand...
We have had open statement that embargo is over on March the 2nd. Plus, the XFR video also states even the hour of embargo lift.
 
ole-m said:
if it's designed for 10 mb L3 and you feed it 20mb L3 and it performs better.. should one disregard it?
just a question about progress in cache size that intel have really rocked for years :)
if some app is made using same algorythm and it greatly benefits from larger cache, it would mean it performs much better? no? but in cpu-z's defence it would not give a overall performance number that the application probably wants.
Anyhow, it just tells us it's not a great datapoint :p
Click to expand...

It is a bit different:

 
CSI_PC said:
I know what Stilt said and also others on other forums that went into more detail, furthermore my point is it does also improve for AMD CPUs but not necessarily with the greatest gains for Ryzen.
Click to expand...

Do you have a link? I vaguely remember what he said, and I would like to check his exact words and context.

razor1 said:
They specifically stated XFR @ 4.1 was on One Core.
Click to expand...

Is there confirmation that XFR is for one core only? The only temperature-dependent frequency on Ryzen is F_TMAX, which is one-core, but a Red Team guy said me that XFR is an all-core turbo. I don't know if trusting him on this or not.

lolfail9001 said:
None of the numbers here reference review sample, however. It is an embargo breach if he does not lie, though. OTOH if the TDP thing is true, i will be pissed off.
Click to expand...

We have known for a while that AMD is shipping "95W" Qualification Samples whose real TDP is higher than 95W.
 
Last edited:
XFR is single core boost, why would it be an all core boost if the max all core boost is 4Ghz.

The knitpicking Ryzen does this because of X and Y is also tedious, this is a good CPU because it was built that way, it is good at Cinebench because it is really good, FX wasn't just deal with it. We are no on to the ultimate cherry pick finding Intel's major advantage in spec INT and proclaiming it important like the second coming of Christ. 99% of people in this market segment have no use for AVX at all.

First it was good at Blender, AMD has hidden single thread behind SMT, Then CPC's leak, not much help but it was really low clocked indicating upliftment, then a bunch of leaks on CB nobody wanted to believe then the reveal that AMD was not hiding anything.I think we need to move on from trying to find inferiority in this CPU, because there is none.
 
Last edited:
Reality said:
doesn't matter, the cheapest AIO cpu cooler will get you an equal OC as a custom loop. been there, done that.
Click to expand...

I wouldn't go that far. The cheapest AIO's can't handle extended loads as well as towers. The liquid eventually heats up. It all boils down to airflow vs surface area. A tiny radiator is a tiny radiator.
 
GroundZero There are custom loops that only use one radiator for the CPU. So of course you'd need to make a direct comparison. Just like a 360mm AIO would slightly outperform a 120mm custom cpu-only loop (when running a stressful application for a prolonged period of time)

But if you compare a Xmm AIO to the same Xmm custom loop, the overclocks aren't going to vary at all (certainly not enough for a max overclock)

When I went from an H100i to my raystorm and 2 radiators, I was pretty unimpressed with what I initially thought would be at least a few degrees difference
 
Reality said:
GroundZero There are custom loops that only use one radiator for the CPU. So of course you'd need to make a direct comparison. Just like a 360mm AIO would slightly outperform a 120mm custom cpu-only loop (when running a stressful application for a prolonged period of time)

But if you compare a Xmm AIO to the same Xmm custom loop, the overclocks aren't going to vary at all (certainly not enough for a max overclock)

When I went from an H100i to my raystorm and 2 radiators, I was pretty unimpressed with what I initially thought would be at least a few degrees difference
Click to expand...

Agreed, IMO the best you can do when you have adequate cooling is to lap both surfaces and apply thermal paste with a straight edge(both surfaces). Tighten as much as is safe.
 
lolfail9001 said:
None of the numbers here reference review sample, however. It is an embargo breach if he does not lie, though. OTOH if the TDP thing is true, i will be pissed off.
Click to expand...
The OCing, sounds like he was using the utility and that has a lot of complex hooks.
Really would prefer them to just do this through BIOS without the app-utility to begin with, but if his posts are true then strange he is ignoring the traditional way.

Those quotes are making the 95W figure interesting and possibly disappointing, maybe that is base clocks only, if he is telling the truth.
But then considering the performance is around Broadwell, reality should also tell us it is unlikely to be 95W at stress test when the 6900K is 140W in that scenario, maybe a more real world stress figure is around 120W.
Cheers
 
lolfail9001 said:
Rumor is on the street that Skylake-X 8 core already beat 6950X record in Cinebench. So, it may be very expensive, but it sure is fast.

If Intel thinks it can get away with $1k octo core again, IT must be tripping.
Click to expand...
It is Intel :)
But more seriously it has IMO a poor CEO now in Brian Krzanich , his approach has not been good for Intel and I doubt we will see a sensible response that retail consumers will be drawn to.
So looks like nice performance shame about the price.
Cheers
 
You must log in or register to reply here.
Back
Top