Digital Foundry Article Technical Discussion [2018]

[BRiT]

PPM (pixels per minute). It’s a stat that’s independent of the developer methods but can also be used to complement them.

But that is still very dependent on how long a gamer is expected to remain in certain areas, which depends on the gamer type. Even if you look at a Multiplayer Map, a single level could have many different values because of how the map plays which depends on the game mode; TeamDeathMatch plays vastly different than King of the Hill or Capture The Flag. Each Map+GameMode combination will also have different sections if the team is going for power weapons or support positions. A single map could easily have small zones (indoor corridors), large far zones (outdoor sniping), and medium zones (transition hot-points or choke points).

Talk about High Complexity with High Effort required with perhaps little to no appreciation for the net result.

 

[Globalisateur]

Some games will need to have those statistics based on level. It was pointed out that the recent Sonic game was patched to have different resource usage based on the level stage, where some busier levels have lower res while less hectic levels have higher res.

Them once you have the stats per level, you need to weight those based on how often and for what duration of time the user is in that level, which means another set of statistics mean, median, 95th and 99th percenticle for the various level of gamers. Like if a gamer rushes through the main story a level could be less than 2% of the game but if they're going for all Achievements/Trophies that same level could be 25% of the game.

But that is still very dependent on how long a gamer is expected to remain in certain areas, which depends on the gamer type. Even if you look at a Multiplayer Map, a single level could have many different values because of how the map plays which depends on the game mode; TeamDeathMatch plays vastly different than King of the Hill or Capture The Flag. Each Map+GameMode combination will also have different sections if the team is going for power weapons or support positions. A single map could easily have small zones (indoor corridors), large far zones (outdoor sniping), and medium zones (transition hot-points or choke points).

The same reasoning could be applied to framerates.

 

[BRiT]

The same reasoning could be applied to framerates.

Yes, completely agree.

Maybe also worth noting if the Res@FPS is during user controllable sequences or during cut-scenes (or other different behavior sections like Tutorial or Introductory Levels), since there's a few engines now with peculiar adjustments to drop down FPS but increase Res during cinematics even if the hardware can easily handle higher FPS at slightly lower Res.

Though some caveats may need to apply if you're running VRR at over 60fps.

Definitely plenty of various trade-offs are made in games which makes for complications in capturing the details and presenting to others if you're an outside party performing the tasks.

 

[bunge]

But that is still very dependent on how long a gamer is expected to remain in certain areas, which depends on the gamer type. Even if you look at a Multiplayer Map, a single level could have many different values because of how the map plays which depends on the game mode; TeamDeathMatch plays vastly different than King of the Hill or Capture The Flag. Each Map+GameMode combination will also have different sections if the team is going for power weapons or support positions. A single map could easily have small zones (indoor corridors), large far zones (outdoor sniping), and medium zones (transition hot-points or choke points).

Talk about High Complexity with High Effort required with perhaps little to no appreciation for the net result.

Different gamers don’t necessarily matter. If we understand the max and min technically of what a game is capable of then gamers will settle into their preferred style and know and they’ll know what to expect.

 

[BRiT]

Meltdown and Spectre: does patching your PC impact game performance?
Security fixes hit some applications hard - but does that include gaming?

http://www.eurogamer.net/articles/d...-cpu-security-flaws-impact-gaming-performance

It's been described as one of the most serious PC security issues we've ever seen. Headlines have been dominated by the news that there's a serious flaw in the hardware design of Intel microprocessors, going back over 20 years and covering millions upon millions of CPUs still in use today. If your gaming PC is based around an Intel processor, it's vulnerable to the recently revealed 'Meltdown' exploit. Side by side with Meltdown is another serious security problem, dubbed 'Spectre' - which has the potential to affect your PC, smartphone or tablet, regardless of whether or not it has Intel technology at its core.

So at the very basic level, what's at risk here? Essentially, Meltdown allows malware to gain access to protected memory within your CPU, areas within your processor that should be impossible to access. Sensitive data of just about any description is potentially accessible. Spectre offers another vector in acquiring sensitive data, to the extent that, although more difficult to deploy, it may well be causing headaches months or even years from now.

In the short term, fixes are being rolled out - and more may yet follow - and the concern is that performance may be significantly impacted. Epic Games recently revealed the extent of the hit for its Fortnite gaming servers, with a revealing graph showing a huge spike in CPU utilisation once the Meltdown patches were installed. The question is to what extent the patches - such as they are now - will impact the average gaming PC.

 

[Nesh]

So Meltdown is specific to Intel, but Spectre affects all CPUs?

 

[iroboto]

So Meltdown is specific to Intel, but Spectre affects all CPUs?

i liked this post to say yes to both answers. But then I realize that wasn't clear.
So
Yes; Yes.

 

[3dilettante]

So Meltdown is specific to Intel, but Spectre affects all CPUs?

Meltdown at a minimum affects the ARM A75 and at least some Apple CPUs.
There are unknowns as to other vendors that have ARM derivatives like Qualcomm, or architectures like IBM's SystemZ and POWER. The statements concerning them are ambiguous.

Various CPUs that do not speculative extensively enough can be immune to both. The Apple Watch is unaffected by either, per Apple.

edit: Posting error made me miss Intel's range and unknowns concerning Nvidia or other ISAs not being often reported on.

 

[turkey]

We need a tool to make stats about resolution: mean, median, 95th and 99th percentile etc.

But first we need a thread for that.

Some games will need to have those statistics based on level. It was pointed out that the recent Sonic game was patched to have different resource usage based on the level stage, where some busier levels have lower res while less hectic levels have higher res.

Them once you have the stats per level, you need to weight those based on how often and for what duration of time the user is in that level, which means another set of statistics mean, median, 95th and 99th percenticle for the various level of gamers. Like if a gamer rushes through the main story a level could be less than 2% of the game but if they're going for all Achievements/Trophies that same level could be 25% of the game.

It seems that the developer has spoken and gone into some depth on how their game approaches this and offers some numbers to how the game scales overall on screen resolution

http://34bigthings.com/redout-a-technical-analysis-on-the-offline-dynamic-resolution-scaler/

it seems like a novel solution given track racing such as theirs is somewhat deterministic on view and load.

 

[Billy Idol]

Cudos to the devs for this article. Nice!

Cool way of doing the scaler...I wonder if they could have used machine learning to train the offline scaling?

If I understand the article right, only 2% of the time the resolution is 1080p?!?!

What the heck DF!!!! What did you measure??

Meh, DF needs better QA if this debacle is really true...this would also explain why the devs went mad.

 

[AlNom]

If I understand the article right, only 2% of the time the resolution is 1080p?!?!

What the heck DF!!!! What did you measure??

Meh, DF needs better QA if this debacle is really true...this would also explain why the devs went mad.

There's a new patch coming, so the data might not correlate between what they present now vs what was measured manually last week.

A Redout Xbox patch is rolling out on January 11th, which will sensibly improve performance for all Xbox users. Announcement of this patch was supposed to come the same day of the release, because that’s the best way to go for a small title like ours, but current things standing, we have no choice but to announce it now. This is optimization work we have concluded months ago. We are currently working on yet another patch, which release date cannot be confirmed yet.

 

[Deleted member 11852]

In the short term, fixes are being rolled out - and more may yet follow - and the concern is that performance may be significantly impacted. Epic Games recently revealed the extent of the hit for its Fortnite gaming servers, with a revealing graph showing a huge spike in CPU utilisation once the Meltdown patches were installed. The question is to what extent the patches - such as they are now - will impact the average gaming PC.

Spectre, where it develops and and the measures potentially required to combat it in the future, and the performance impact from those measures, could be considerable. It's too early to say. The shadow of Spectre will likely be long.

There are unknowns as to other vendors that have ARM derivatives like Qualcomm, or architectures like IBM's SystemZ and POWER. The statements concerning them are ambiguous.

Other vendors are talking to customers direct, this has been going on for a while.

 

[3dilettante]

Other vendors are talking to customers direct, this has been going on for a while.

I did find Nvidia's updates and advisories for various Jetson and Tegra products.
While I don't think I've found a clear indication for the first Denver due to Nvidia's having different variants of the Tegra K1 with ARM versus Denver, there's something of a game of omission for the Tegra products with respect to Meltdown that might mean that it is.
http://nvidia.custhelp.com/app/answers/detail/a_id/4616

The Jetson TX2 uses Denver 2, and Nvidia indicates there are mitigations for meltdown.
http://nvidia.custhelp.com/app/answers/detail/a_id/4617

 

[Deleted member 11852]

I did find Nvidia's updates and advisories for various Jetson and Tegra products.

From where I sit, in my ivory tower, it looks like the public communication is limited to the companies who directly or indirectly sell to the public. Everybody else, is talking but not publicly. Frankly, it's an easier conversation when both sides understand the issue and you don't have to worry about the media confusing the issue or misreporting things, but I think ARM's comms were ace. No bullshit, no legal bullet dodging, just what you want and need to know

 

[3dilettante]

From where I sit, in my ivory tower, it looks like the public communication is limited to the companies who directly or indirectly sell to the public. Everybody else, is talking but not publicly. Frankly, it's an easier conversation when both sides understand the issue and you don't have to worry about the media confusing the issue or misreporting things, but I think ARM's comms were ace. No bullshit, no legal bullet dodging, just what you want and need to know

Per this, IBM has confirmed Meltdown and Spectre for Power, although for me the links don't currently work.
http://www.zdnet.com/article/meltdo...mware-and-os-fixes-for-vulnerable-power-cpus/
System Z is still unclear.

For IBM, the first rumblings were from Redhat, where the open-source communication channel cannot be closed. Other rumors are that a chunk of this uncertainty is related to IBM's scrambling to respond.

 

[Deleted member 11852]

Per this, IBM has confirmed Meltdown and Spectre for Power, although for me the links don't currently work.

https://www.ibm.com/blogs/psirt/potential-cpu-security-issue/ works for me.

 

[3dilettante]

https://www.ibm.com/blogs/psirt/potential-cpu-security-issue/ works for me.

For whatever reason it doesn't work where I am currently.

 

[Deleted member 11852]

For whatever reason it doesn't work where I am currently.

 

[BRiT]

That dev article was being ripped apart in the ResetEra thread yesterday. I have not had a chance to see what they state explicitly or how things have faired today.

 

[3dilettante]

I was able to view the page later, and after some digging around I did find an entry for a Power 8 firmware documenting fixes for all three exploits. I didn't dig any further, so I am taking the news and Redhat statements indicating this extends to Power 7.
There's something incoming for Power 9, and research is underway for prior Power generations.
System Z is not something as readily browsed for.