Guild Wars 2

Ryzen 1700 @ 4.00GHz
R9 290
1080p

This is for the people who are curious about the new Ryzen chips.

Seems on par with what I expect. 8 cores/16 threads doesn’t matter for a game that has arguably 2-ish CPU intensive threads. Improved IPS over Bulldozer is a huge plus.

Do you have reflections on? Everything else seems to be on high I just can’t tell.

Behellagh.1468:

Seems on par with what I expect. 8 cores/16 threads doesn’t matter for a game that has arguably 2-ish CPU intensive threads. Improved IPS over Bulldozer is a huge plus.

Do you have reflections on? Everything else seems to be on high I just can’t tell.

yeah everything is maxed except shadows, I remember if you max out shadows bad stuff happens so I just set it to high.

Looks like Ryzen isn’t that bad at all…especially if you are using the lowest end 7 series. To hold 15+ FPS without stuttering shows a lot of potential.

New video with more information:

Try benchmarking LA in primetime (just stand on a few spots and look around, look at the bank, crafting stations, mystic forge, etc). Next to a full on 70vs70vs70 in SM, LA is generally what stresses the machine the most.

Thanks for the video. What program displays all thise stats?

ArchonWing.9480:

Thanks for the video. What program displays all thise stats?

MSI Afterburner, Its a great program but sometimes it can conflict with GPU drivers in certain cases. Other than that it’s a great tool a lot of people use.

Behellagh.1468:

Seems on par with what I expect. 8 cores/16 threads doesn’t matter for a game that has arguably 2-ish CPU intensive threads. Improved IPS over Bulldozer is a huge plus.

Do you have reflections on? Everything else seems to be on high I just can’t tell.

The difference between the FX series and Ryzen is huge. In the pvp lobby I would get around 45fps in the center where everyone crowded. If I began to run towards the golem training area I could get 60fps (no higher) since not many players were being displayed. Note my FX was OCed to 4.6ghz always.

With the Ryzen 1800x, I get around 65-75+ fps in the crowded area and when I’m running towards the golems I get around 120fps.

PS: Running at 2560×1080 with fov maxed, the fov maxed reduces the fps by a ton. If I put it at default, the fps almost double.

So we know the Ryzen 1700 performs well, and the 1800x even better. I’m betting once the SMT issues are fixed to allow the Cores proper access to their Threads we may see even better performance.

Right now I’m watching and debating…1700 or 1700x + ASRock x370 Taichi + 8-16G DDR4 and my RX470 should be a decent match up for the next 4-5 years…hopefully at least.

moonstarmac.4603:

So we know the Ryzen 1700 performs well, and the 1800x even better. I’m betting once the SMT issues are fixed to allow the Cores proper access to their Threads we may see even better performance.

Right now I’m watching and debating…1700 or 1700x + ASRock x370 Taichi + 8-16G DDR4 and my RX470 should be a decent match up for the next 4-5 years…hopefully at least.

For the added cost and marginal mhz boosts, the 1800X and 1700X are completely pointless CPUs. You only need to OC a 1700 by 300mhz to reach 1800X level performance – and 1800X is unlikey to reach any higher with the 4ghz wall (unless you win the silicon lottery). Just bumping it a mere 100 or 200mhz to reach like 95% of the performance should be pretty much guaranteed.

That’s an OS scheduler fix so it’s in Microsoft’s court. The 1700 is $170 cheaper than the 1800X and may be overclockable to a similar clockspeed, which makes it the value choice of the three.

Charder.5730:

ArchonWing.9480:

Thanks for the video. What program displays all thise stats?

MSI Afterburner, Its a great program but sometimes it can conflict with GPU drivers in certain cases. Other than that it’s a great tool a lot of people use.

Thanks! I tried it out and it’s quite fun.

As a suggestion, I would recommend changing the stat colors to white or something, as the default color is really hard to see. It’s accessible in the rivatune icon in the system tray.

Also, I’d like to see a video using more ‘reasonable" settings that you would normally play at, since it’s already proven that some settings (character model/quality) isn’t designed for any computer. Like how, high can you crank it up and maintain high fps in crowded situations?

like she said the software needs to update to make it work with AMD cpu. Microsoft is coming out with update for win10.

I’m very much looking forward to the 1600x after reading up on it today. Sure you lose 2 cores and 2 threads, but it should be about on par with the 1700 in the long run.

I am waiting though until the Rev2 / updated bios boards hit the market and the Windows 10 Scheduler Update is live.

Yea the 1600x is the one people should look seriously at for this game. The motherboards look pretty affordable too.

As a side note, is the i7-k processor really better than the i5-k processor for Gw2? I would imagine it runs into the same issue of those threads not being used. On my locked 4 thread i5-6500, I can never get Gw2 to use more than 75% so it seems that 3 cores is all that it really wants.

I was very curious, and it turns out that AMD don’t think that Windows 10 scheduler changes are either needed, or helpful: https://community.amd.com/community/gaming/blog/2017/03/13/amd-ryzen-community-update?sf62107357=1

Behellagh.1468:

That’s an OS scheduler fix so it’s in Microsoft’s court. The 1700 is $170 cheaper than the 1800X and may be overclockable to a similar clockspeed, which makes it the value choice of the three.

Kinda the same back then with the 8120 vs the 8350 which was just an oc’d version of the 8120 yet i was able to oc it to 4.8 ( not very stable ) then turned it down to 4.6 stable in most games and found it’s sweetspot to be 4.4.

SlippyCheeze.5483:

I was very curious, and it turns out that AMD don’t think that Windows 10 scheduler changes are either needed, or helpful: https://community.amd.com/community/gaming/blog/2017/03/13/amd-ryzen-community-update?sf62107357=1

The issue isn’t SMT per se but the scheduler moving a thread from one CCX quad to the other as L3 cache is local per CCX and and memory access between the two is at the same speed as from main memory.

Behellagh.1468:

SlippyCheeze.5483:

I was very curious, and it turns out that AMD don’t think that Windows 10 scheduler changes are either needed, or helpful: https://community.amd.com/community/gaming/blog/2017/03/13/amd-ryzen-community-update?sf62107357=1

The issue isn’t SMT per se but the scheduler moving a thread from one CCX quad to the other as L3 cache is local per CCX and and memory access between the two is at the same speed as from main memory.

Yes. …and that is what AMD specifically addressed, and said was not something that a Windows scheduler fix would be needed, or helpful, for.

I mean, it could be that AMD are wrong about the way the CPU their built performs, but it seems … less likely.

No, Microsoft could fix it by treating the Ryzen the same way as two physical CPUs in the system, a setup often found on server class motherboards. Once a thread is assigned to a CCX, it can’t be reassigned to the other CCX and can only move between the cores of the CCX. Windows 10 does support 2 physical CPUs.

The SMT thread scheduling “problem” that doesn’t need fixing is the old assign threads first to unused cores first and only then start assigning a second thread to a busy cores.

The original AMD FX Bulldozer scheduler issue was it double up threads on modules first to reduce the number of modules used so power reduction of the unused modules would allow clock boosting of the busy modules. The issue was the clock boost couldn’t compensate for the performance loss caused by both “cores” running within a single module fighting over the shared core components within the module, which IIRC was on the order of 10-15%.

Part 2 (had to get the grocer before it closed —-—-

The issue of the OS resigning an active thread from one CCX to another is independent of SMT as the issue would still exist with SMT disabled.

Now AMD may not think the issue is fixable because there is no way for the OS to distinguished the two CCX quads from each other when it enumerated cores so the scheduler would treat them as two 4 core CPUs ver one 8 core. And this is important as the R3 and R5 quad and hex core CPUs are still using the same 8 core die with two or four cores disabled. And that may lead to hex cores with 3 working cores on each CCX or 2 on one and 4 on the other. The quad core Ryzen may end up with 2+2; 1 + 3 or 0+4 working core arrangements with some performing noticeably better than others. That’ll make the “CPU Lottery” impact more than just about OC headroom.

For now, Ryzen users would be wise to experiement with manually setting the affinity when doing intensive tasks such as gaming.
Read my sig, I’m using Process Lasso, a superb tool for just about everything cpu related – highly recommended.

Well just found out R5 is due out April 11.

Behellagh.1468:

No, Microsoft could fix it by treating the Ryzen the same way as two physical CPUs in the system, a setup often found on server class motherboards. Once a thread is assigned to a CCX, it can’t be reassigned to the other CCX and can only move between the cores of the CCX. Windows 10 does support 2 physical CPUs.

I’m sorry, I really don’t follow you here. Are you saying that AMD are wrong, and have failed to understand thread scheduling on their own hardware?

Thread Scheduling
We have investigated reports alleging incorrect thread scheduling on the AMD Ryzen™ processor. Based on our findings, AMD believes that the Windows® 10 thread scheduler is operating properly for “Zen,” and we do not presently believe there is an issue with the scheduler adversely utilizing the logical and physical configurations of the architecture.

As an extension of this investigation, we have also reviewed topology logs generated by the Sysinternals Coreinfo utility. We have determined that an outdated version of the application was responsible for originating the incorrect topology data that has been widely reported in the media. Coreinfo v3.31 (or later) will produce the correct results.

https://community.amd.com/community/gaming/blog/2017/03/13/amd-ryzen-community-update?sf62107357=1

…though, I guess you are being consistent with the posters there, who also respond that AMD don’t know what they are talking about, and people out on the Internet are better at understanding the AMD CPU than AMD themselves…

If the Ryzen can’t reveal which core is part of which CCX to the OS, then there isn’t any fix. Your choice as a company is to either say their isn’t anything that can be done or say it’s say it’s working as expected. Which answer would you choose?

OP, can you like your ram speed

https://www.reddit.com/r/Amd/comments/5zr8lv/i_asked_amd_a_followup_question_about_infinity/?utm_content=title&utm_medium=hot&utm_source=reddit&utm_name=Amd

rumor has it that infinity fabric is tied to ram speed

The Infinity Fabric connects CCX to uncore devices such as memory controllers, PCIe, SATA, USB etc. It is a 256 bi directional crossbar and runs at the speed of the memory controller ( ie when using 2133 Mhz memory, the memory controller clock speed of 1066Mhz is the clock speed of the fabric. Using faster memory increases fabric speed )

edit nvm, listed in you vid. 3200 Mhz

Behellagh.1468:

If the Ryzen can’t reveal which core is part of which CCX to the OS, then there isn’t any fix. Your choice as a company is to either say their isn’t anything that can be done or say it’s say it’s working as expected. Which answer would you choose?

Yeah, I’m sure you are right. It’s not like there is any way that Windows understands multiple processors at the level you are talking about.

…whatever, though, honestly. It’ll come out in the wash.

ArchonWing.9480:

As a side note, is the i7-k processor really better than the i5-k processor for Gw2? I would imagine it runs into the same issue of those threads not being used. On my locked 4 thread i5-6500, I can never get Gw2 to use more than 75% so it seems that 3 cores is all that it really wants.

No one answered this, but the answer is no. You might see a slight increase because you can dedicate OS/other program resources to threads/cores not being used by GW2 but clock for clock an i5 and an i7 will get you the same performance in GW2.

I went from a 4.5GHz i5-3570k to my current 4.5GHz i7-3770k and there was 0 difference in GW2.

Fermi.2409:

ArchonWing.9480:

As a side note, is the i7-k processor really better than the i5-k processor for Gw2? I would imagine it runs into the same issue of those threads not being used. On my locked 4 thread i5-6500, I can never get Gw2 to use more than 75% so it seems that 3 cores is all that it really wants.

No one answered this, but the answer is no. You might see a slight increase because you can dedicate OS/other program resources to threads/cores not being used by GW2 but clock for clock an i5 and an i7 will get you the same performance in GW2.

I went from a 4.5GHz i5-3570k to my current 4.5GHz i7-3770k and there was 0 difference in GW2.

Thanks. I wanted to know because since this game is cpu heavy people tend to throw around “get an i7” which is already vague but now I know this info is misleading.

While we’re at it, do you think the new i3s and those hyperthreaded g4560s would do fine in Gw2? I know only 2 real cores would hurt but single core would still be strong. In particular, in comparison to someone that has an fx 8350.

I know a friend that built a cheap i3 system and I will probaly do some experiments

I had an i3-3220 before the i5 and it was all right. It’s hard to compare to the i5 since I upgraded my GPU and CPU At the same time, both of which had/would have had an impact at that point. Generally it should be fine though if you’re not watching videos or anything when you play, especially so if you do some overclocking

SlippyCheeze.5483:

Behellagh.1468:

If the Ryzen can’t reveal which core is part of which CCX to the OS, then there isn’t any fix. Your choice as a company is to either say their isn’t anything that can be done or say it’s say it’s working as expected. Which answer would you choose?

Yeah, I’m sure you are right. It’s not like there is any way that Windows understands multiple processors at the level you are talking about.

…whatever, though, honestly. It’ll come out in the wash.

Now you are just being difficult. And all that does is allow a user process to choose which logical processors that application can run on.

I’m talking about OS scheduler awareness so every thread started doesn’t wander out of it’s CCX as it moves between logical cores during the course of it’s life. But I can see in some cases that may also be as bad as needing to access the other CCX’s L3 cache across the “infinity fabric” bus in other cases.

And my concern over oddly arranged hex and quad core Ryzens has been dispelled as AMD at the R5 announcement assured people that the hex core will only be in a 3+3 configuration and the quad 2+2 . Still unclear if the R3 will be a new die with a single CCX or just more lemonade making (or brain damaging) of the same die the rest of the Ryzen series uses.