Hi all. I've been playing with a little idea to make a Benchmark style test with GG. It's not perfect, has an occasional mouse issue , but otherwise it seems pretty good.
I include a demo here for people to try out. If I can get myself interested I may go a bit further with this, I'm a bit of a perfectionist and was really waiting to get it 100%, but you may never get to try it in that case.

Let me know what scores you get I think it should be fairly accurate, but with only one machine to test on hard to be 100%

Edit - I forgot to disable the low performance message, so many are getting this annoying popup. To remove and get the best result you can, change hidelowfpswarning=0 to hidelowfpswarning=1 in the standalones setup.ini before running.

I've added an updated lua script that fixes the mouse not locking. I've also added the setup.ini to stop the low fps message annoyance for your convenience.



https://drive.google.com/open?id=0B4qIchrlXIZrYWF5N2c4UmFuamc

Ok i tested although i did get a "Low FPS" warning despite the fact it seemed smooth as silk ?

RPGMARK : 3431.0193558506

That is down to Game Guru, and the fact I missed changing the ini file to disable it It will be running slower than you think, but as it's non interactive it looks better. Good score though, actually higher than I got on my new rig. I assume it ran okay? No weird mouse issues forcing you to view the floor? There's an occasional bug which causes this to happen and of course the score would be significantly higher, as it is normally looking directly at the ground.

I also thought I'd floored the score ;/

Hi DVader, very cool I also got the low FPS warning

RPGMark: 3068.2712264099

Yes, if you go into the setup.ini you need to change hidelowfpswarning=0 to hidelowfpswarning=1.

This will stop the low performance warning and also give you a more accurate score as the message popping in will skew the results.

Glad you like it

Tried it but the mouse issues forcing you to view the floor happens,tried it 3 times.also ran as administrator to see if it would help with no luck.

Dave

Ran fine as far as I can tell no mouse problem
Score 3851.0840801686

haven't tested it as i'm afraid my slow internet would take too long to download it but if you're using the SetFreezeAngle(g_PlayerAngX..) etc then you need to wrap the x angle like this

I got 5293.9663908762 no mouse issues and no low fps warning

6268.

Crappy little i5 with an AMD GFX card.

Reminds me at WoW a little bit ^^.

3794.*

how are you guys getting such high scores, I changed the hidelowfpswarning=1 but my score is low for my system, I assume my GFX is poop
I am running 3 monitors from this GFX card could that cause it?

Both your graphics card and your processor are much lower than ours.

To start with, you have an old i7 (2000 series) i have the latest i7 (7000 series), and i assume Amens i5 is at least a 6000 series, there is a huge difference in performance between them and yours.

As to graphics a big mistake people make with NVidia is thinking the higher the number the better the card, this isn't always the case.

The first number, in your case 9, just tells you what series the card is, basically how new the card is, unless there are major changes in the technology on a new series then it's only of secondary importance. The number you really want to take note of is the second part, in your case the 50, this tells you what type of graphics card it is, this number usually goes up in stages from 10 to 80 and is very important to take note of when choosing.

This is a very rough guideline, it's not always the case, but it's close enough:

x10 to x40 are generally mainly for web browsing, office work, and very very low end games
x50 is for very low to low end gaming
x60 is for low to medium end gaming
x70 is for medium to high end gaming
x80 is for high to extreme gaming

Because there is two numbers to think about when choosing a card it can be a little confusing, because you would expect a 950 to be better than a 780 because it's a newer card, but in reality a 780 actually beats a 950 at most tasks, obviously the bigger the gap between the series the less the difference is.

A 950 is actually the equivalent of a 760 or a 660ti, a 960 is roughly the equivalent of a 670 or a 580, and likewise a 970 is actually pretty close in speed and power to a 1060.

This is also a major reason people can't play some games very well, and why they flip out because they think they meet the requirements, a game might say "minimums graphics 780" and people think "ooh i've got a 950, my cards much better!" then suddenly after buying the game they say "hold on why am i only getting 20fps?", and flip out at the publisher, what they don't realize is that they don't actually meet the requirements at all, they just don't understand how the numbers work. That's one of the reasons when people say "i meet the requirements" on tech reports we always ask them to list their specs, because sometimes they think they meet the requirements, but actually don't.

and yes, using three monitors on a graphics card can cause its performance to lower.

lol sheesh thanks for the explanation new kit ordered will be here tomorrow

Haha. It's still a good card though, just not as powerful as some think. This is why i'm a big fan of benchmarks, and very happy DVader came up with this. If companies stopped recommending by card and gave minimum and recommended benchmarks instead there would be a lot less irate customers out there.

803 ...... on the built-in Intel 4600 . Runtime: a cool 30 Minutes. [ 805 without the low fps panel being shown]
2965 ...... on the Nvidia 780M GTX
(with the low fps warning panel shown and clicked away)
1920x1080

Edit:
3512 on the other notebook with Nvidia 980M GTX (and a little higer resolution )

BTW: does anyone know whether the s2 port is hardwired to the Nvidia, usually ? I am getting the same results regardless of choice of integrated Intel GFX or built-in NVIDIA, but the monitor is run via s2 port (to get the 4K reesolution if I need it

Yep my i5 is pretty new, and water cooled to boot, and 32G of super fast DDR4 RAM. Gfx card is a Nano, don't know why they dropped off the radar as fast as they did as they are terrific little cards.

Of course I cheated to get such a high number @ such a high resolution. Real figures at that res were 3880 and about 4500 at 1080p.

I'd try it on my i7 extreme system but my son has borrowed it.

DVader, great work, this is almost exactly what I was thinking of when I mentioned the usefulness of a GG Benchmark in the other thread.
Hope that many will post their results and specs to give others the chance to assess their machines' GG capabilities

Thanks everyone. It seems from results so far it works pretty well to test your system for how well it will run Game Guru with a reasonable sized map and a few AI

@Smallg. Not looked into the mouse issue as yet but it seems odd it only does it occasionally. The x angle shouldn't ever change from 0 really as that is what I am forcing it to with the freeze command. I'll have to look into it when I get chance. The code is a bit rough and ready really but does the job (mostly).

@ amenmoses. 6268. Hmm, I thought that sounded a bit high. I imagine you altered the script or the setup defaults I didn't bother trying to pack it securely or anything

@yrkoon. It was your post that inspired me to try this. I wanted to test the latest version out but had nothing in mind, so I thought I might as well see if my idea worked okay I've done a couple of cut-scene angles with this sort of thing, but never tried actually moving around.

Lol, nah didn't edit anything, simply knocked the mouse at the beginning so it was looking at the floor.

I got 5472 on my desktop and 2652 or something there about on my laptop. It tends to take over mouse control, however, and not give it back. It ran me into the lake on the laptop, and I drowned. Still got a benchmark, though, the progress bar was pretty near the end. Is it possible I lost points for getting killed? Just kidding, aside from the mouse issues (also forces you to look at the ground a lot; I don't know if that's intentional, for comparison reasons, or what), it's a pretty useful utility, thanks!


EDIT: I know there are a LOT of entities on that level, especially trees. Any idea how many?

I got 3970. I ran into a few problems. I walked away from my pc to get a drink. I came back and I was stuck in a wall. I managed to get back on track. The mouse kept wanting to point down at the ground. But I did make it to the end. I'm going to re-run it when I have more time. This time I won't walk away Lol. My Specs-Cpu- amd 8350,8 gigs of ram,49 inch tv monitor @ 1080p,graphics card- rx 480,1 TB standard HDD, Running windows 10 64bit. Thank You DVader for the benchmarker! It's just what I needed.

My fastest PC, windows 10, AMD Radeon HD 6700M.

RPGMark: 923.99299613309

I watched the RED status saying SLOW PC for around 30 min.

@ Jerry Tremble. There are 1420 objects on the map, so it's a reasonable test. I want to add a bit more scenery to make it more interesting as it rolls along yet though, so it may get a few more added down the line. I may have to check the store for something to put on the water that fits the theme. I have loads of media, but boats seem thin on the ground.

@ Preben. 30 mins, wow that's certainly a long run!

I have sat down and looked at the mouse problem and have finally fixed it. I'll post the new script at the top of the thread. That should get it running correctly now with no wandering off track and looking at the floor

How are you calculating the final score? What is it based on?

6658 with mouse problem (pc in signature) with normal hdd no ssd
1920x1080 no ovc ddr3 1600mhz

One important stat people should be reporting is the screen resolution you are running at! The numbers you get at VGA are going to be way different to those at 4K.

Less meaningful are hard drive stats, all that will effect is the loading time which isn't being measured.

Might also be useful to mention the mobo chipset if you know it and RAM stats, i.e. DDR3/4 and clock speed, single/dual/quad channel.

For example my 3880 figure is with i5-6500/3.2GHz, 32GB DDR4 Dual channel (2Ghz), 2560x1440 AMD-Nano (no overclocking). Z170MX - Gaming 5 chipset.

I've just realised my EZTune setting is currently in the preset ECO mode! I wondered why the system is so quiet.

Cyclecount/time taken. Basic but should be fairly robust.

I'm currently adding a little more to it and I'll update the upload when it's done

I ran the test and got 3465.xxxxx

i5 - 4670k CPU @3.40ghz
Nvidia 660gtx
8gb RAM
64bit Windows 7

My PC is 4yrs old now and out of date . Been thinking about getting a laptop .

Just curious.
Is the code in script? Could we see the code?

I tried the test with ssd, and without touching the mouse.

Score : 5448.
Msi z97, i7 4770k (no ovc)
Nvidia gtx 1070
16gb ram ddr3 1600mhz
SSD 850Evo 250gb
1920x1080 resolution

Yes it is. No other place to put it really
@ Lance, the score you got seems fairly good to me. I got a similar score on my brand new Ryzen 1700 system, which isn't particularly out of date

@ Lotgd, don't forget to update the lua file and setup.ini with the new ones I've put up at the start of the thread. The mouse should be completely fixed in that version. Good score by the way.

I'm a little closer to a fresh update, which will be a little more demanding than this version as I'm adding in some more interesting stuff along the path. I've already dropped my own score by a fair few points.

DVader, you are on such a good way that I dare to ask:
would it be a lot of work to you to put out a simple, line by line report with some more basics,
such as
the mark value
processor type detected,
Ram detected ,
gfx adapter GG thinks it is talking to,
running time of the test
screen resolution used ?

That way, a sticky thread could be generated, people could be told to run the test first, then post their report file,
thus presenting their results in a consistent and comparable way without further instructions.

Those who think they are suffering from performance problems could be directed to first compare their individual machine's results to the reports of machines of a similar class ( in terms of gfx card and cpu, in the first place)

If they are in a +/- 15% channel with results already posted for their class. they'll know where they are (and maybe, it helps them to see they are not alone, there). And it should reduce the discussions on "... but I can run this AAA game and that at a much higher framerate."

And those definitelyy outside of this 15% channel really may have a reason to complain and ask to be helped, I'd say.

I've looked into it a little, but it seems there is no way from lua to get these types of stats. I can get the resolution, which I have already in the updated version yet to be released, but nothing else beyond the running time, which should be possible. I'd need Lee to add lua commands for CPU, GPU etc. Should be fairly easy for him as they are obviously available via the F11 key

Saving the results is certainly possible if I have enough info worth the bother

I could also go mad and add multiple or optional tests I imagine, from a top end test down to lowest settings possible. That is if I could set the quality levels down in lua which at the moment is not possible It could still be done, but only by changing draw distance and such, rather than shader settings. I also have collision enabled on anything that should have it in a game, so could disable that for a speedier version as well. I left it in so it is a better measurement of gaming performance.

I'd need Lee to add lua commands for CPU, GPU etc. Should be fairly easy for him as they are obviously available via the F11 key

Saving the results is certainly possible if I have enough info worth the bother

Yes, I agree to both statements, actually, I was thinking that the F11 data were available already.

But: I know I originally also had voiced my thoughts on more variations in the other thread, but after seeing your first version of RPGMark, I readily dismissed them as probably too much trouble / for insufficient ROI on effort to result.

There is probably simply too much variation of too many parameters to play them all through, and there is still the number of non-static objects and and and ... not controlled by parameters, beyond that.

I think If a user can be shown that his gear produces similar results as other, similar gear with ONE set of settings , that's enough, and he can then be pointed to a set of hints how to improve performance by reducing certain settings. He'll need to sort out differing effects on his particular game, anyway.

My mark was 3116.xxxxx. My machine is now over 6 years old, but I have a GTX 1050, i7 Sandybridge original chip.
Thanks for posting the test, scenery looks great.

@yrkoon,
I agree, this will give people an indication of how GG will run on their system. It's obviously going to struggle on lower end systems as it's deliberately trying to stress the system as a benchmark normally does I have in fact left out a few things which would drag it's speed down further, but as they don't really affect the look that much, felt they shouldn't be implemented. Better to have more scenery and such than effects which slow it with no real visible effect.

@JimmyC, 3116 is still a decent score from my tests and the results here. Lee sent me a score of 5600 on his dev machine so that is probably not too far off as high as you will get on a normal (air cooled, non clocked) machine. The results will really depend massively on your video card more so than the CPU here, the CPU will only drop the score (by a lot) if it is a really old one, like my old Q6600 or similar.

Of course the new versions scores will be slightly different/lower as I have added in more scenery to the level. I've run into some odd bugs with some new entities I have put in, which seem to be no longer working in the preview version. I would have released it by now otherwise.

Hummm. Am I missing something? I ran it and after some 10 minutes or so the bar below showed me the message "Fast PC". No SCORE so far...

Hitting ESC brought me to the menu.

Mate. HOW the heck do I get the score to show up?

btw: I got the low pc message and could click it away. No probs there.

EDIT: No mouse-problems so far.

Please me! /.-)

You wait until it is finished and it shows it in the middle of the screen, at full speed and high fps I think it took about 15 minutes to get to the end.

Thanx Belidos,

And I thought I am on the patient side of humans /.-)

Yeah it is a bit long. The trouble is that I designed the path control first and made a path to go round. That was fairly quick, but once I added all the scenery, the 2 min test took a little longer The latest version tested on mine at 6 and a half mins or so on my last standalone run, interestingly running directly in GG, the same level took 9 mins, so I imagine there is quite some difference with standalone speed there.

You will get ridiculously long run times on slow computers though. Still it serves it's purpose, on that PC you will struggle to use GG for any large levels.

@DVader

And that is very well. To hell with benchmarks deliberately betraying their users.
When it is like you stated, the user will probably get better reading marks on some of his level and worse on others, so, he will get the impression that the benchmark is "right" on average, and that is the best reliable result he can get in my opinion.

Ok, now I let it run through. My benchmark is humble, but I encountered no problems what so ever and everything ran smoothly... Supernice level btw DVader! I took a stroll...

2387
[With my Phenom II X4 set to each 3900MHz and my Palit GTX 750 Ti [passive] set to 1146 MHz GPU / 2934 MHz RAM Clock]

Screen resoultion: 1680 x 1050 [2 Screens]

BONUS-question: Would anything in the 2500er ballpark be a good performance for a "typical target PC"-development-system? Regarding the fact that most users don't have high-end-systems?

Cheers again!


Does anyone with experience here know what an update from a

- Phenom X4 (4 real cores with CPU / FPU running @3900)

to an

- AMD FX-8350 (8 CPU cores / 4 FPU cores running @4000)

would probably mean in increase regarding this benchmark here? I'm taking intuitive guesses as well

I have an FX6200 and my score with a GTX960 is 24xx

Dvader, very nicely done

Hard to say until there are lots of results in really, but that score seems reasonable. I'm pretty close to getting a new version finished which will probably be the final version. It is a little more demanding, but I have also upped the camera's speed to make the test runs a little faster or at least no slower I seem to have found an odd issue with the latest public beta release though with the players start angle. There has been an update since I tried it yesterday, so that may be fixed now.



I doubt there would be a massive difference with that cpu, as it's only 100mhz faster overall and GG's multi-threading probably doesn't extend to using all 8 cores. I could be wrong though.

Thanks Earthling It's been a bit of a labour of love as I have had to sit through the test many, many times while making it :0 Should be a decent little Benchmark test in the end though.

Thanx DVader. I took a look at some benchmarks. There are gains of 30% to 50% in speed increase in single- and multi-core-tests. So using a FX8350 could score some decent 2800 with my graphic-card-setup... Well worth an upgrade, since the processor is only ca. 140,- EUR in germany now...

You should be _ok_ with that, because I feel _fine_ @ 2900, although my Intel I7 CPU runs at a speed way lower than your AMD, it hardly ever scratches the 3.4GHz (from below) ; look at the wonders Preben is crafting with his <1000 RPG Mark as a developer.

If you want to make a living out of it and GG is your development target, it is obviously best to get the fastest machine you can afford; if you are a hobby developer "only" setting out to make THE GAME(s) you always dreamt of and never could get elsewhere, then it won't make so much of a difference if an intermediate generation will take 8 minutes as opposed to 5 ; as far as I know, even "monster machines" won't compile any non-trivial GG level in only seconds.

It seems the latest public build has possibly broken my map I've reverted back to the last version because of a camera issue in the newest public preview. The camera is working okay again, but the map seems to have developed a green glitch in the distance that is pretty noticeable. I've tried deleting the superpalette.ter file in the lush directory, but it has made no odds. I've verified Game Guru with no luck either. I'm hoping it is a simple fix, as I do not fancy reworking the level from an older version as it has many changes I've not had the heart to even look how far back it will set me.

Edit - Reverting back to the latest version solves the green glitch issue, but puts me back to the buggy camera issue The joys of programming I've no idea why it's starting in the wrong direction in this version, but simply changing to it affected the camera's initial direction.

That's one of the biggest problems comparing Intel with AMD, people like it straight forward and like to compare numbers, but they just don't compare, an Intel processor has a shorter pipeline, so it takes less processing power to complete a task, but with AMD processors the pipeline is longer so it takes more power to carry out the same task. It's a bit like driving from your house to your work, you take the shorter route it uses less fuel, but if you take the long route it uses more fuel.