AMD Ryzen (Summit Ridge) Benchmarks Thread (use new thread)

[CentroX]

Intel engineers calls zen competitive

https://www.techpowerup.com/230446/amds-ryzen-chips-10-smaller-than-comparable-intel-skylake-dies

 

[KTE]

You're using a rather dated Z97 Haswell setup at a low clock rate of 4.3Ghz though, it cannot compare with a Skylake/Kabylake at 4.8-5.1Ghz.

Which is supposedly faster than top Ryzen.

You said it yourself.

Sent from HTC 10
(Opinions are own)

 

[Magic Hate Ball]

What % of 4790Ks could even get to 4.9GHz? What % of 7700Ks can get to 5GHz+?

Cheapest X99 board: https://www.newegg.com/Product/Product.aspx?Item=N82E16813157542&cm_re=x99-_-13-157-542-_-Product - $190

Cheapest Z270 board: https://www.newegg.com/Product/Product.aspx?Item=N82E16813128975&cm_re=z270-_-13-128-975-_-Product - $110.

You are right, there is a significant price differential here, but the difference isn't inherently due to the socket or the chipset, it is due to the fact that X99 boards are explicitly targeted at enthusiasts. They need to have higher quality components because the mobo makers need to assume that they are going to be overclocked, and the assumption is that performance/features matter.

Just as some examples, the cheap-o X99 board supports SLI, while the cheap-o Z270 board does not. You get an Intel LAN chip on the X99, but you get a Realtek on the Z270, etc.

Really, I don't know why somebody would cheap out on the motherboard for a serious gaming PC.

With Ryzen incorporating more features onto the SoC instead of the motherboard, prices should be even cheaper than Z270, and far cheaper than X99 motherboards.

For pure gaming bang for buck, a minimalist motherboard that allows for 1-2 SSD's (and/or 1-2 HDD's) and one 1x16 PCI-E 3.0 slot should be the perfect gaming setup. SLI/Crossfire are a compatibility liability, and going for a single more powerful GPU is the best answer.

The Ryzen B350 motherboards should be perfect for 95% of gamers, as it has overclocking capability but no SLI/Crossfire. This cheaper platform also will support anywhere from 4c/4t (speculation) to 8c/16t... it's the perfect low cost entry platform with the greatest range of upgradability.

How is it a bad idea for a person who's not going to SLI/Crossfire, use a 6 disk RAID array, or have a need for a 10gigabit ethernet controller???

Anything past that are garnish for someone who just wants to GAME.

 

[DrMrLordX]

According to Siliconlottery.com 7700K's require higher voltage to reach 5.0 then 4790K's take to reach 4.9. So 100Mhz improvement for slightly higher power draw..

1). Skylake/Kabylake have higher IPC than Haswell (especially when paired with proper RAM). So you get more than "100 MHz improvement".

2). Intel's 14nm+ has different leakage characteristics than Intel's 22nm process. Current draw is not the same. So an i7- 7700k at x voltage does not draw the same power as an i7-4790k at the same voltage. I would dare say that the 7700k would probably draw less average over any given number of samples.

3). Very few Haswell samples - even among Devil's Canyon chips - hit those clocks anyway. You were normally lucky to get 4.8 GHz, with 4.6 GHz being the more common.

 

[krumme]

That's what I'm saying to him too.

I was playing BF1 today on a 64 player map. My core utilization on my 8 logical cores (i7 4770k @ 4.3ghz) hit 90-10% very often for all 8 threads. We're already pushing the limits in multiplayer especially.

This. On mp 64, operations, saint quentins scar some sectors and situations i forecast i will be able to even break a 8c 4.3 skl cpu. Its horrendous cpu load when 64 man 2 planes and 2 tank battle for a single building in one big destruction - but damn fun

 

[bjt2]

Intel engineers calls zen competitive

https://www.techpowerup.com/230446/amds-ryzen-chips-10-smaller-than-comparable-intel-skylake-dies

15% lower capacitance and lower operating voltages means something, especially on 4c chips...
Given that infos I don't think that 4c Zen will have LOWER clocks than top BR APU (4.1/4.3)...

 

[DeeJayBump]

Intel appears to be playing the role of 1990 Mike Tyson, while AMD appears to be playing the role of Buster Douglas. All indications are that Intel is about to be crawling around on the mat looking for it's mouthpiece, which is somehow fitting with all the mouthpieces inundating the forums at the moment.

EVERY champ or seemingly invincible titan can be beaten or toppled by a younger, hungrier competitor.

 

[CHADBOGA]

Would you buy a 7700K for $350 when a 6C Ryzen is available for $350 and performs within 10% worst case?

In what applications will a 6c Ryzen with a base clock of 3.4 to 3.6Ghz, be within 10% of a 7700K with a base clock of 4.2Ghz?

How many threads will this 6c Ryzen have?

 

[beginner99]

I kept my 875K for 6 years and would have kept it longer if a friend hadn't given me a huge deal on a i7 4770K + mobo + RAM combo...

Lol. I'm still running with that one. But I also don't always needed the newest games. In fact waiting some time saves money and nerves (not being alpha-tester). I will go 7700k, Ryzen, or Skylake-X.

7700k vs 6-core Ryzen has one small advantage. The iGPU. First one is that in case dGPU fails you can still actually use your PC while waiting for RMA/getting a new one. Assume you are 1 month from a new GPU release. Do you go 1 month without PC or buy basically an outdated GPU?
Second point I can argue is that not only could games use more cores, they could actually make use of the iGPU be it for physics or dx12 multi-adapter tech. Demos of this have already been shown and offer roughly 10% increase with intel GT2 graphics. So there are arguments on both sides, not just cores when talking about longevity. And of course if this actually becomes reality, Ryzen APUs with HBM would be king...

 

[Tuna-Fish]

^^ why are AMD stocks rising so fast it is becouse of Ryzen if it is how does unreleased and untested CPU rises stocks so high?

Stock price is basically investors prediction of the future profits of a company. Any news that would increase profits in the future naturally raises the stock price. Being about to release a product that is much more competitive than the company's past products obviously qualifies.

 

[CentroX]

According to this http://www.linleygroup.com/mpr/article.php?id=11666 analyst. Ryzen is on par with Ivy bridge and even lags behind haswell

 

[dfk7677]

According to this http://www.linleygroup.com/mpr/article.php?id=11666 analyst. Ryzen is on par with Ivy bridge and even lags behind haswell

That is only a calculation based on +40% compared to Excavator.

First, we recalculated the A10-7850K’s benchmark score without libquantum, which ICC has cracked, resulting in an adjusted score of 81.4. Increasing that number by 15% for an Excavator-based design should yield 93.6, and a 40% boost from moving to Zen yields 131.

 

[Atari2600]

According to this http://www.linleygroup.com/mpr/article.php?id=11666 analyst. Ryzen is on par with Ivy bridge and even lags behind haswell

That is recycled from David Kanter, who based it on a 40% improvement on Excavator.

AMD have already claimed to have exceeded this 40% (with other people estimating it to actually be up to 55%), so the graph is somewhat null and void.

 

[CentroX]

Yeah i guess it should be 55% now.

 

[lolfail9001]

Figure 2 shows our estimates for Excavator and Zen. First, we recalculated the A10-7850K’s benchmark score without libquantum, which ICC has cracked, resulting in an adjusted score of 81.4. Increasing that number by 15% for an Excavator-based design should yield 93.6, and a 40% boost from moving to Zen yields 131. We further expect that using a compiler optimized for Zen (instead of Intel’s compiler) would boost performance by another 10% to 144.

Sorry folks, but it is calculated on:

1) Assuming Excavator is 15% faster than Steamroller

2) Assuming Zen is actually 54% faster than Excavator (144/93.6~1.54).

So, if anything it is actually using later figure that AMD have provided.

 

[inf64]

Sorry folks, but it is calculated on:

1) Assuming Excavator is 15% faster than Steamroller

2) Assuming Zen is actually 54% faster than Excavator (144/93.6~1.54).

So, if anything it is actually using later figure that AMD have provided.

Actually Kanter claims in the article that the figure refers to Integer spec_rate benchmark which is wrong.
We already have ballpark figure for IPC, look no further than CPC preview. They measured 35% higher throughput for 1C/2T Zen vs 1M/2T PD, at the same clock.

It shouldn't be that hard for most of you to work out how much IPC for ST that is, knowing that average 2T CMT gain for PD is ~1.66x and that SMT gain for Zen should be between 1.2 and 1.25x. To save you the math, according to CPC's findings, IPC gain Vs PD should be around 74-80%, or 52-56% Vs XV core (depending on the SMT/throughput gain Zen has when running 2T on a single core).

 

[SAAA]

Sorry folks, but it is calculated on:

1) Assuming Excavator is 15% faster than Steamroller

2) Assuming Zen is actually 54% faster than Excavator (144/93.6~1.54).

So, if anything it is actually using later figure that AMD have provided.

Even better: it was made months ago and fits today 55% IPC estimations, and considering Excavator isn't 15% faster than Kaveri in all situations I'd even call it an over-estimation, for spec int of course. Other benches do show Zen beating Broadwell but that's not on average I guess.

 

[crashtech]

We're now reduced to looking at articles from last August?

 

[CentroX]

We're now reduced to looking at articles from last August?

Architecture is still the same.

 

[Shivansps]

Lower IPC and lower clocks is not really a good match 8C/16T Ryzen@4Ghz looks very good vs the ridiculus overpriced 6900K, it can match it.

But, les take for example a 3.7Ghz 6C/12T, its likely slower than 6800K, and its gona have a really hard time vs the 4.5Ghz 7700K, except on some especific heavy MT tasks

The 4C/8T. gona depend fully on HT when compared to an I5, that is not going to give a clear winner.

turbo clocks of 6C and 4C needs to be higher than the turbo clock for the 8C.

 

[sm625]

Intel appears to be playing the role of 1990 Mike Tyson, while AMD appears to be playing the role of Buster Douglas. All indications are that Intel is about to be crawling around on the mat looking for it's mouthpiece, which is somehow fitting with all the mouthpieces inundating the forums at the moment. .

In other words, IPC increases?

 

[LTC8K6]

We're now reduced to looking at articles from last August?

And using Mike Tyson as an example of...something...

 

[.vodka]

We're now reduced to looking at articles from last August?

Sadly yes. An article based on information we had before New Horizon with the 40% figure being the only thing to take into consideration. You know, people seem to have forgotten about those verifiable demos AMD set up in that event clearly showing >40% increase in relevant metrics with the CEO stating such a thing.

In other words, IPC increases?

No JF-AMD spreading FUD he was fed from within this time around, it's different.

Amazingly, IPC has actually increased this time, seemingly enough... 3 weeks to go.

 

[coercitiv]

7700k vs 6-core Ryzen has one small advantage. The iGPU. First one is that in case dGPU fails you can still actually use your PC while waiting for RMA/getting a new one. Assume you are 1 month from a new GPU release. Do you go 1 month without PC or buy basically an outdated GPU?

How can you establish which system actually has the upper hand when you have no clear idea for a Ryzen 6c system cost?

 

[Doom2pro]

I think one of the aspects lost on all this good Zen news is, considering they used a lot of Bulldozer knowledge in the design, Bulldozer is no longer a complete failure...

The name Zen is for a reason, they took the best of both high frequency and high IPC and created Zen, so the question has to be asked... Would AMD be where they are now (with Zen) if they hadn't put all the years of engineering into Bulldozer to get all that high frequency IP, low FO4 as bjt2 likes to put it?