AMD Ryzen 7 1700X 8 Core CPU Benchmarks Leaked Thousand Dollar Performer At $389

As expected, IPC is close to broadwell but AVX adoption is Sandy/Ivy level, since this is an enthusiast part AVX is hardly important. Great to see IPC is way up in single thread performance and Floating point ops.
 
I wouldn't make any conclusion yet. Its an awful bench.
upload_2017-2-11_16-30-32.png
 
Shintai said:
I wouldn't make any conclusion yet. Its an awful bench.
Click to expand...
Still validates both AMD Demo's, early Geekbench leaks and recent rumors of 5Ghz perf matching 5Ghz 5960X.

Frankly, the curious part now is what is up with memory performance. That could legitimately ruin it as gaming CPU, if it stays on retail samples.
 
lolfail9001 said:
Still validates both AMD Demo's, early Geekbench leaks and recent rumors of 5Ghz perf matching 5Ghz 5960X.

Frankly, the curious part now is what is up with memory performance. That could legitimately ruin it as gaming CPU, if it stays on retail samples.
Click to expand...

Source in RoC said the IMC is very limited and will be updated by Zen+, I guess it is baby steps but overall it is a positive outlook.
 
Also as WhyCry says the AMD benchmark result is potentially crippled a bit due to the budget motherboard and also those memory timings are pretty bad, wondering why such memory/motherboard would be use for such a leak: (17-17-17-39 2T) @ 2400 MHz.
If true not bad results tbh but some question marks about the setup IMO and worth considering these are synthetic tests.
Really need to see it with a decent board and memory and various applications-tools that can analyse single-multi core performance, and of course gaming.
Cheers
 
Last edited:
jOtTCB.jpg


Entry level motherboard and some horrific timings on the RAM is a bit of a gimp job.
 
CSI_PC said:
Also as WhyCry says the AMD benchmark result is potentially crippled a bit due to the budget motherboard and also those memory timings are pretty bad, wondering why such memory would be use for such a leak: (17-17-17-39 2T) @ 2400 MHz.
If true not bad results tbh but some question marks about the setup IMO and worth considering these are synthetic tests.
Really need to see it with a decent board and memory and various applications-tools that can analyse single-multi core performance, and of course gaming.
Cheers
Click to expand...

Yeah test performance on a motherboard that doesn't support beyond 2133 along with other skimpings, maybe we can run a 7700K on a H110 lol
 
CSI_PC said:
Also as WhyCry says the AMD benchmark result is potentially crippled a bit due to the budget motherboard and also those memory timings are pretty bad, wondering why such memory/motherboard would be use for such a leak: (17-17-17-39 2T) @ 2400 MHz.
Click to expand...
These DDR4 timings are close to JEDEC standard for DDR4-2400. Such mobo was probably used because test was done by pc builder (google the reported name), so these obviously try to cut corners on mobo cost. And looking over at computeruniverse hints me that 2400c17 sticks are the cheapest.
Do note, though, that timings alone do not even come close to explaining the horrific memory performance. Either mobo/BIOS is bad, or Zen IMC is.

Anyways, i am fairly excited, looks like a decent cheap workstation CPU to get, if ECC will be available on mobos.
 
lolfail9001 said:
These DDR4 timings are close to JEDEC standard for DDR4-2400. Such mobo was probably used because test was done by pc builder (google the reported name), so these obviously try to cut corners on mobo cost. And looking over at computeruniverse hints me that 2400c17 sticks are the cheapest.
Do note, though, that timings alone do not even come close to explaining the horrific memory performance. Either mobo/BIOS is bad, or Zen IMC is.

Anyways, i am fairly excited, looks like a decent cheap workstation CPU to get, if ECC will be available on mobos.
Click to expand...
Yeah why I also said budget motherboard.
I am still not entirely convinced with that reason for why they leaked with such timings, because that would be more important to server/workstation segment and then you would not leak on such a motherboard.

And yeah agree it looks positive for AMD on the CPU side, just a shame the performance leak is ruined by the setup, if it is real data.
The setup reminds me more of a non-professional setup and test.
Cheers
 
CSI_PC said:
I am still not entirely convinced with that reason for why they leaked with such timings, because that would be more important to server/workstation segment and then you would not leak on such a motherboard.
Click to expand...
Because that is the cheapest mobo and memory money can buy, duh. It is basically the realistic representation of what OEMs will do with it.
Snowdog said:
Quite decent except for Physics and Primes. Any idea what happens there?
Click to expand...
Current running theory is that these use AVX2 somehow, and Ryzen has half the throughput of Haswell/Broadwell/Skylake in these. Needs verification of some sort, however, because otherwise it is not clear at all.
 
I'm looking to upgrade for VR gaming, I need high minimum framerates. I think this leak puts me towards an Intel build. I just ran the test on my sons amd 860k 3.8ghz -- stock heatsink fan / settings /etc in single threaded I scored 1745 to that ryzen 2046. Meanwhile I3-4160 scores 2104 -- If they make an i5-7640k with hyper threading for $249 I'm going that route...
 
Physics is SSE based.

But again, the benchmarks are very tiny and if you can loop it or not got a big impact.

And before people mention the INT performance. Remember this:
upload_2017-2-11_18-37-52.png


You better wait for more proper benchmarks.
 
just looking at Integer math graphs, my 3.4ghz fx-8320e is beating a 4ghz 6700k by 10%... so I don't trust these at all.
 
Yeah, some of those results seem weird, to say the least. That benchmark seems pretty inconsistent. Still, seems like Ryzen is in the ball park of being pretty competitive.
 
buttons said:
I'm looking to upgrade for VR gaming, I need high minimum framerates. I think this leak puts me towards an Intel build. I just ran the test on my sons amd 860k 3.8ghz -- stock heatsink fan / settings /etc in single threaded I scored 1745 to that ryzen 2046. Meanwhile I3-4160 scores 2104 -- If they make an i5-7640k with hyper threading for $249 I'm going that route...
Click to expand...
So, $50 here and $100 there ends up adding up, but that having been said, the 7700k is $350 msrp, and you can find it on sale for $300, so it's not the craziest idea to get an i7 which we know is quad-core hyper-threading (vs the unsubstantiated 7640k rumors). But Ryzen is coming out soon anyways, so all AMD questions will be answered then.
 
DAFUQ @ the naysayers here citing RAM speeds as game crippling.

2000 and its DDR2 called. It wants to know what games you play that aren't almost completely GPU bottlenecked, even moreso with adaptive sync implementations.

Running 4K just mighty fine @ 2133 for the past 3 years
 
The low result for ryzen in primes surely is because this is a AVX2 benchmark. Also in this bench the memory bandwith must be very important because the intel i7-7700k has also a very low result (the i7-700k has 2 memory channels, the others intel parts have 4 memory channels).
 
Manuel Jose said:
Also in this bench the memory bandwith must be very important because the intel i7-7700k has also a very low result (the i7-700k has 2 memory channels, the others intel parts have 4 memory channels).
Click to expand...
Or because 7700k has half the cores of HEDT parts, as can be evident from looking at 5820k result that sits exactly in the middle of 6700k and 6900k (+/- few percent).
 
In the physics benchmark it happen same similar. This benchmarks must be memory bandwidth limited. This is typical in this type of benchmarks.
 
lolfail9001 said:
Or because 7700k has half the cores of HEDT parts, as can be evident from looking at 5820k result that sits exactly in the middle of 6700k and 6900k (+/- few percent).
Click to expand...

If the result for i7-7700k is because this processor is only 4C/8T versus the other intel parts, the value in this prime benchmark should be approximately 40, not 27.
 
Manuel Jose said:
If the result for i7-7700k is because this processor is only 4C/8T versus the other intel parts, the value in this prime benchmark should be approximately 40, not 27.
Click to expand...
I have no clue why it was 27, but 6700k@4Ghz is 31, stock 5820k is 49 and 4Ghz 6900k is 77. Down to minor margins it scales linearly with frequency and core count on these 3.
Manuel Jose said:
In the physics benchmark it happen same similar. This benchmarks must be memory bandwidth limited. This is typical in this type of benchmarks.
Click to expand...
Does not apply to physics either.
 
lolfail9001 said:
Because that is the cheapest mobo and memory money can buy, duh. It is basically the realistic representation of what OEMs will do with it.
Click to expand...

Unfortunately it does not work that way if this was from a certified related provider/builder, they would use a model specifically designed for that model range.
Could you imagine a Xeon or HEDT being presented with the lowest rung consumer board in terms of performance benchmarks from Intel?
Answer is no because they are not designed for them.
From my experience anyway but everyone has a different situation and experience.

I checked the AMD slides and Ryzen 8C+ is specifically meant for at minimum Mainstream board and that is B350 not the A320.
The A320 cannot even support the boost clocks (which aligns with the test not showing such clocks), reduced USB offerings, maybe no PCIe x4 (it seems I assume if using PCIe x16 dGPU or was an old spec slide from AMD), amongst other limitations.
Seriously wrong motherboard even for workstations (unless low entry-basic use ones and this then has to then consider the CPU as well) and is really the bottom rung board when excluding small form factor, along with the unusual memory says it all IMO for that leak test and if not a fake pretty good results considering.

Remember this test/'leak' was done with one just down from the top Ryzen CPU (competing with and aligned to I7 and HEDT), not the budget or more mainstream CPUs and crucially APUs such as Bristol Ridge, context here that it should had been with a more 'correct' motherboard and its chipset/power stage/memory used/etc.
Cheers
 
Last edited:
lolfail9001 said:
I have no clue why it was 27, but 6700k@4Ghz is 31, stock 5820k is 49 and 4Ghz 6900k is 77. Down to minor margins it scales linearly with frequency and core count on these 3.

Does not apply to physics either.
Click to expand...
If a 6700k@4Ghz has a value of 31 for this prime benchmark, why the 7700K one is 27 (a lower value) with a faster clock?
 
Couple of important slides when considering the motherboard:
From 2016:
Bristol-Ridge3.png-nggid048378-ngg0dyn-800x0x100-00f0w010c010r110f110r010t010.png


From 2017:
am4_05.png


Part of why I feel the test result if it is not fake reduced the results or at least caused some of the inconsistencies we see, and an unusual combination of entry motherboard-memory and 'HEDT' equivalent CPU.
Pretty good results considering the quirks (and if not fake), but they are synthetic and as mentioned really need a deeper analysis of applications including games and tools to look at single and multi-core performance and bottlenecks.

Cheers
 
Last edited:
There are too many quirks in the benchmarks for me to place much stock in them, but I see a few possible bright spots. It's obvious AMD has truly made a big jump in performance with Ryzen, and I for one am happy to see this. Increased competition is a good thing for enthusiasts and consumers.

I'm not a die hard fanboy of either brand, but I'm hoping we get a true contender from AMD. I just bought my 6600k last year, so I doubt I will be upgrading anytime soon, but I would love to have a 2nd viable option to offer to people I occasionally build systems for.

Hopefully the platform itself offers ECC on consumer gear and doesn't grow stale like AM3+ eventually did. Call me spoiled, but I've grown to enjoy high performance and platform longevity.
 
Gman1979 said:
There are too many quirks in the benchmarks for me to place much stock in them, but I see a few possible bright spots. It's obvious AMD has truly made a big jump in performance with Ryzen, and I for one am happy to see this. Increased competition is a good thing for enthusiasts and consumers.

I'm not a die hard fanboy of either brand, but I'm hoping we get a true contender from AMD. I just bought my 6600k last year, so I doubt I will be upgrading anytime soon, but I would love to have a 2nd viable option to offer to people I occasionally build systems for.

Hopefully the platform itself offers ECC on consumer gear and doesn't grow stale like AM3+ eventually did. Call me spoiled, but I've grown to enjoy high performance and platform longevity.
Click to expand...

Yep, they run in seconds and some dont even use L3. And the workset for example for INT and FP is so small that they can run in a 512KB cache or bigger. And in those benches you would also see some very odd results with for example SKL-X/EP vs SKL/KBL due to 1MB L2 vs 256KB L2.
 
m7z8yr.png


0FJ6Q20.png


So it looks like they benched AMD on a entry level board, with turbo disabled from BIOS likely due to stability and failed to disclose that much.
 
OrangeKhrush said:
So it looks like they benched AMD on a entry level board, with turbo disabled from BIOS likely due to stability and failed to disclose that much.
Click to expand...

The A320 boards do not support overclocking and not designed for the high end 8C CPU, so possibly firmware with compatibility for such a tier board or cannot even do Boost totally correct and separately potential issue with phases-power demand.
Might also explain some of the inconsistencies beyond memory, disappointing they did not use a B350 board and reasonable memory, which is why I am wary of this test.
Cheers
 
Last edited:
CSI_PC said:
Also as WhyCry says the AMD benchmark result is potentially crippled a bit due to the budget motherboard and also those memory timings are pretty bad, wondering why such memory/motherboard would be use for such a leak: (17-17-17-39 2T) @ 2400 MHz.
If true not bad results tbh but some question marks about the setup IMO and worth considering these are synthetic tests.
Really need to see it with a decent board and memory and various applications-tools that can analyse single-multi core performance, and of course gaming.
Cheers
Click to expand...
is it a rhetorical question or are you trolling, you answered your own question, to cripple performance in purpose.
 
Deathroned said:
is it a rhetorical question or are you trolling, you answered your own question, to cripple performance in purpose.
Click to expand...
Err why is that trolling?
I am actually defending the AMD CPU for now because the results due to how the anonymous person tested/setup is pretty unreliable to compare to Intel I7 benchmark/results lol :)
And yes of course the question is food for thought that it is a compromised test (btw I NEVER said it was deliberately compromised so no idea where you got that from), as I mentioned in other posts I seriously doubt this has anything directly to do with AMD or even their certified partners/builders.
So considering all of that the results are pretty good (my posts are around the use of a A320 budget 'essentials' board with the poor timing memory with an 8C HEDT equivalent CPU and AMD suggest mainstream boards minimum).


Cheers
 
Last edited:
Deathroned said:
you answered your own question, to cripple performance in purpose.
Click to expand...
Look at name of Passmark account. It is a firm that at the glance sells pre-built PCs. A320 mobo and DDR4C17 are the cheapest components. It perfectly checks out.
 
lolfail9001 said:
Look at name of Passmark account. It is a firm that at the glance sells pre-built PCs. A320 mobo and DDR4C17 are the cheapest components. It perfectly checks out.
Click to expand...
The problem is AMD slides in the past has suggested these should be used with the mainstream motherboard and not the more budget essentials.
As I said before, could you imagine a 6c/12 or 8c/16 Intel (putting aside this is not technically possible) with the cheapest range of consumer motherboards and with some of the poorest timing memory available?
The A320-motherboard range is more for APUs and possibly the lowest Ryzen CPU models.

It is a compromised pre-built PC IMO considering the CPU used.
I would need to check but way back HP was suggesting a solution of A320 with Bristol Ridge APU for their OEM.
Cheers

Edit:
Yep as an example the OEM for HP is/was 65W supporting 3 APUs: http://support.hp.com/ca-en/document/c05254568
So OEM will also look towards the Raven Ridge going forward on the A320 board IMO.
I appreciate that will probably also include the lowest Ryzen models as well.
 
Last edited:
You must log in or register to reply here.
Back
Top