AMD Ryzen 5 3600 Review Leaks, Shows Impressive Performance

[HD64G]

Good to see the 2700x still holding up very well against a newer generation of AMD cpu's.

Not in single-threaded apps and in multithreaded ones it goes well only vs the weakest of the Ryzen 3X00 CPUs which will cost $200. I like 2700X but Zen2 arch is on another level on all aspects (efficiency, clocks, cache, etc).

 

[efikkan]

Time to move on from DDR4-3200? Wonder if DDR4-3600/3700 would improve write bandwidth, and latency?

Higher memory speed will potentially improve bandwidth (if there are no other bottlenecks), but not latency.
Time to move on? These are the first products rated for DDR-3200, and it's plenty for most uses.

10% in RAM latency and 3200 which affects IF frequency, not the best conditions. It looks it isn't even CL14.

I'd like info on the older AMD and Intel setups to compare apples to apples. 3200 is low even for Zen+ IMHO.

What is wrong with testing the product under the maximum rated speed?
Running beyond 3200 MHz is overclocking and is no longer a benchmark of the product at stock.

 

[Casecutter]

I read it as the Intel has the AIO. In the x470 review the Wraith was mentioned but I didn't see it mentioned in the x570 review.

Okay, I see in the X470 it says that "we have used the reference heatsink" in the Temperature and Consumption section. While in the X570 article there's a block of text under the Test Equipment that said "Equipment used for this comparison with the Intel Core i9-9900K".

Now we don't know which version of Wraith, but as the R5 2600 and this is a 65W part it probably the Wraith Spire. So as a Core i9-9900K doesn't include a stock cooler we have no idea what Intel would have to actually limit that chip to if they had to box it with some prohibitive lub of aluminium. See if they said this is all the cooler/cost we can stomach to give would they still be able to infer the same performance/clocks/TDP at only about $500? That's what I'm getting at, the Core i9-9900K is great because they can un-hinder it from any base required cooling system. If AMD said this part not held to 65W, say more like the Core i9-9900K and you find you own cooling, what would AMD be able to bin parts at and what would that be able to offer? Sure the way AMD provides the clocks/TDP it's copacetic with the Wraith Spire we believe they will provide. CPU spec's in many respects are tethered to what the cost (and total price point) that a cooling solution can support.

Edit: I have two 2700Xs and they don't start to lose all core boost before start/mid 70 C so if the same is the case here then the stock cooler is sufficient/fair for a review.

Are both running with a completely different Wraith Prism? As that a better and more costly cooler, even demanding tests it will do decently in permitting a 2700X provide it's rated stock configuration.

For a supposed $200 CPU +Cooler to present what is said here is just crazy competitive.

 

[Manu_PT]

Why Should AMD change your mind when You're of Top AMD haters in entire Forum ?

Wanting high framerates/performance is being AMD hater? I also hate Opel because they failed to deliver an electric car with 700km capacity. Low performance, so I bought Audi instead. Because "I hate" Opel.

This is your pathetic logic.

If only you'd have a clue about how these things work you'd realize that if AMD was able to increase IPC even though memory bandwidth and latency remained constant that meant this wasn't the primary constraint.

But keep making stuff up for your template Intel fanboy comments and entertain us.

If you knew how CCX works and how certain engines starve for low latencies instead of raw instructions per clock, you would understand. Keep getting salty bud, try again on 7nm+, maybe then intel is done. For now, nop, not yet.

 

[Fabel]

What is wrong with testing the product under the maximum rated speed?
Running beyond 3200 MHz is overclocking and is no longer a benchmark of the product at stock.

Nothing wrong with that as long as the specs are listed, but neither the CL nor the specs of the other systems are specified.
Intel's superior memory controllers usually allow higher memory clocks and If that is the case on those charts I'd like to know.

What is wrong with asking for all the specs?

 

[Midland Dog]

lol this is either a joke or your delusional.

Zen 2 has better ST than Haswell.

new mobo, new cpu and new ram worth of ST gain? or just buy a proven 4790k that can do 4.9ghz for less

 

[Fabel]

new mobo, new cpu and new ram worth of ST gain? or just buy a proven 4790k that can do 4.9ghz for less

Wait a couple weeks, look at the reviews and decide, but for god's sake, don't buy a CPU that becomes slower every few months thanks to all the security patches.

 

[Midland Dog]

Wait a couple weeks, look at the reviews and decide, but for god's sake, don't buy a CPU that becomes slower every few months thanks to all the security patches.

i haven noticed any slowdowns, probs coz im on an old patch. Is broadwell affected too, coz a 5775c works in my board too

 

[mstenholm]

Okay, I see in the X470 it says that "we have used the reference heatsink" in the Temperature and Consumption section. While in the X570 article there's a block of text under the Test Equipment that said "Equipment used for this comparison with the Intel Core i9-9900K".

Are both running with a completely different Wraith Prism? As that a better and more costly cooler, even demanding tests it will do decently in permitting a 2700X provide it's rated stock configuration.

For a supposed $200 CPU +Cooler to present what is said here is just crazy competitive.

Just wait for a proper review. My equipment is under System Spec.

 

[coozie78]

Exactly what was the point of this ' review '? They're comparing a 8c/16t part that'll set you back over £600 once a decent cooler is factored in with a sub £300 6c12t part. What's worse is the fact the R5 3600 has been hampered with slow memory AND the games chosen are all known to favour either Intel arch or high core counts. Surely an i5 9600/9600K would have made a fairer comparison?

Those latency numbers aren't too great, though, and the memory performance doesn't seem to bode well for those of us who regularly shift fairly large files. Add in the expected high cost of the new '570 motherboards and the requirement for fast-read expensive-memory and AMD may find this a more difficult sell than previous Ryzen releases, particularly to those on older Intel hardware that are looking for an upgrade.

Looks like my 2700X is still safe but, as has already been said by many: Add your own salt and wait until more comprehensive reviews are in.

 

[Vya Domus]

If you knew how CCX works and how certain engines starve for low latencies instead of raw instructions per clock, you would understand.

One simple way to dismiss this mind bogglingly stupid theory that game engines "starve" for latency is the fact that the 3600 is equal or slightly ahead of the 6700K which is reported as having a massive 40% lower latency advantage. Let's see with what sort of made up groundbreaking explanation you come up for this one.

And for the record so that you can learn something new today, IPC is a function of latency among other things. That means that tuning for performance is a multi variable optimization problem where some things can become the bottleneck as you progress through this process of finding the optimum. If AMD increased performance that means latency isn't the bottleneck, plain and simple fact.

Games, as with any other software needs instruction throughput , you can't get better latency without also increasing the instructions throughput and that's what matters.

 

[bug]

Wanting high framerates/performance is being AMD hater? I also hate Opel because they failed to deliver an electric car with 700km capacity. Low performance, so I bought Audi instead. Because "I hate" Opel.

Don't like electric from Opel, buy from the people that brought you dieselgate. You rock, dude

 

[Manu_PT]

Don't like electric from Opel, buy from the people that brought you dieselgate. You rock, dude

Why you using Windows and Google? Want me to describe their dirty measures in the last 10 years? If we go by that we dont use anything!

 

[biffzinker]

Time to move on? These are the first products rated for DDR-3200, and it's plenty for most uses.

I was referring to enthusiasts that buy overclocked DDR4 that's above JEDEC spec. AMD had a press slide with DDR4-3600 as a cost/performance sweet spot or DDR4-3700 as a high performance/lowest latency, and runs at a 1:1 ratio with Infinity Fabric.

 

[Steevo]

One simple way to dismiss this mind bogglingly stupid theory that game engines "starve" for latency is the fact that the 3600 is equal or slightly ahead of the 6700K which is reported as having a massive 40% lower latency advantage. Let's see with what sort of made up groundbreaking explanation you come up for this one.

And for the record so that you can learn something new today, IPC is a function of latency among other things. That means that tuning for performance is a multi variable optimization problem where some things can become the bottleneck as you progress through this process of finding the optimum. If AMD increased performance that means latency isn't the bottleneck, plain and simple fact.

Games, as with any other software needs instruction throughput , you can't get better latency without also increasing the instructions throughput and that's what matters.

IPC has a lot to do with out if order performance with things like gaming where the effect is additive, between thread sync, input, and frame render time it does add up. It's why AMD has added more cache, which limits clock speed and consumes a lot of energy without extra logic to gate more areas, which increases die size and complexity.

No one is saying it's a bad design, but the oddities of the design let us know it's limits, and I'm going to guess that 5.2Ghz is going to be the high end of this design, it will double the power consumption to achieve it on good chips and it's going to be temperature sensitive and we see some death from the variable thermal expansion and two dies soldered in a couple years.

 

[GLD]

I am really happy with my 2nd gen Ryzen, and am really looking forward to Ryzen 3rd gen. My B450 board list support for HyperX 3466MHz, and I bought some and it works great, @1.2v with my Ryzen 2600, sweet!

So surely 3rd gen Ryzen will happily run ram higher then 3200MHz.

Good times ahead!

 

[ToxicTaZ]

AMD 3600X 6 cores will be slower than a Intel 8086K 6 cores cpu.

8086K will remain the world's fastest 6 cores dual channel cpu for 2019.

But the 3600X will have a great price and good power efficiency.

Same thing with AMD 3800X 8 cores will be slower than Intel 9900KS 8 cores dual channel CPU for 2019. But better on price and power efficiency.

So take your pick....price and power efficiency or the fastest 6 or 8 cores CPU.

 

[Xzibit]

AMD 3600X 6 cores will be slower than a Intel 8086K 6 cores cpu.

8086K will remain the world's fastest 6 cores dual channel cpu for 2019.

But the 3600X will have a great price and good power efficiency.

Same thing with AMD 3800X 8 cores will be slower than Intel 9900KS 8 cores dual channel CPU for 2019. But better on price and power efficiency.

So take your pick....price and power efficiency or the fastest 6 or 8 cores CPU.

Doubt anyone be willing to buy 8086Ks at current prices. They are above $700 USD. For that amount you can probably get the 3600X with a X570 MB a SSD and a 16gb 3600 C16 kit.

Even if it was at "MSRP" of $425 for the 8086K you can get a 3600X with a 16gb 3600 C16 Kit and have money left over to go see a movie and buy the BIG JUMBO popcorn with drink.

 

[geon2k2]

impressive no way, i want better ST than intel, no point upgrading from haswell still

According to the El Chapuzas site R5 3600 has better single thread performance than i7 8700K, at least in wPrime.

This should be quite a lot better than Haswell.

 

[Midland Dog]

According to the El Chapuzas site R5 3600 has better single thread performance than i7 8700K, at least in wPrime.

This should be quite a lot better than Haswell.

literally less than %10 difference in IPC from haswell to skylake

 

[Bjorn_Of_Iceland]

AMD gaming people holding out since Phenom, welcome to 2015 I guess?

 

[HwGeek]

I am interested who can test the new Ryzen with power limit of 3~3.5W per core to see what's the max frequency it can operate, since this will give us some info on the new 64C TR that will be 250W TDP for backwards compatibility[50W+ will go for the IO chiplet]

 

[garie234]

I wonder what clockspeeds the 3600 was actually hitting in FC5 it seems to utilize 6 threads from what core utilization benchmarks show so I'm certain that the 3600 is running at < 4.2Ghz maybe 4Ghz or less and the 9900K should be running at ~4.8ghz to 5ghz depending on the Motherboard settings. Based on this the IPC of Ryzen 3x00 is equal to or slightly higher than the 9900k in FC5 if it is running at 5ghz which is likely.

Judging by the cinebench r15 score the 9900k is just running at stock clocks a 5ghz 9900k will score 2150-2200 cb score.

 

[Nkd]

Meh... Skylake 2.0

Also they retested it with x570 motherboard: https://elchapuzasinformatico.com/2019/06/amd-ryzen-5-3600-x570-review/

Amd must match skylake and more cores I will buy.

Amd matches skylake and delivers more cores.

meh, its skylake 2.,0 lol.

 

[efikkan]

Amd must match skylake and more cores I will buy.

Amd matches skylake and delivers more cores.

meh, its skylake 2.,0 lol.

In all the hype it's easy to forget that AMD is catching up to Skylake's performance level. The competition brings the price per performance and price per core down, but performance per core have been largely unchanged for years.

Whether it's worth buying or not depends on what you already have. If you have a Skylake family CPU and your usage don't really benefit from more cores, then wait for the next performance level from either company (or your needs to change). E.g. if you own an i7-8700K and only really do gaming, there is no point in "sidegrading".