Country CUP - 10th Anniversary design thread:

[yosarianilives]

1 hour ago, mickulty said:

I don't know if this is intentional or an oversight but previous competitions have had some stupid petty arguments over the validity of rig photos.  Bit pointless to have to deal with people complaining that one person has in good faith posted a photo from the same session but technically at a moment with a different CPU under the pot (and this happens to be revealed by a cpu-z window on the screen), but if someone was actually faking a photo for some bad faith reason it would be very easy to fake.

It's nice to take and post photos but probably not much use to try and enforce them.

I would agree, I think the best reason for a rig pic is more so everyone can see your cool setup and frosty pics. As a source of discouraging cheating not so much, we've seen already where rig pics encourage people to think there is score sharing when it's really just a group bench session.

[Leeghoofd]

No setup pic required, but always fun to see them right?

Now IF you do a team session plz add piccies so people know what is going on... Also if the moderators suspect sharing you better have the pictures or other proof at hand.

Edited November 4, 2019 by Leeghoofd

[TASOS]

Which exact vga's are capable of running , stage 14 ?

I wanna see , if i have time for a fund-raise ?

?

 

[yosarianilives]

6 minutes ago, TASOS said:

Which exact vga's are capable of running , stage 14 ?

I wanna see , if i have time for a fund-raise ?

?

 

Anything that supports dx11 should be able to run the bench, titans have been explicitly confirmed and workstation cards haven't been shot down any of the times I asked so big bois like radeon pro duo fiji and vega should also be allowed if those are the cards you're looking to acquire  Other than that the cards 100% confirmed to be allowed that should also score well are 7990, titan z, 295x2, 690.

Hope that kind of answers your question

[Strunkenbold]

On 10/31/2019 at 11:15 PM, Leeghoofd said:

Nope moved them to PCIe 3.0(CPU)

On 10/31/2019 at 8:52 PM, yosarianilives said:

For the igp stage is hades canyon allowed? I noticed that in the db Vega M GL and GH are both still listed as integrated, and the cpu used is kabylake so unless they were explicitly denied they would be allowed.

 

I'm honestly not really happy with that move. Depending on who you ask, the definition of IGPs are sometimes GPU integrated in the CPU die or / and CPU and GPU die on the same package. So the Polaris GPU is on the same package with the kaby lake CPU. They are connected via a pcie bus, but it shouldn't be too surprising that they don't use something super fancy given the fact that a Intel part talks to an AMD part. 

Anyway I remember clarkdale having a GPU and CPU die. They were on the same package. Would anyone argue that it's not an IGP?  So I fail to see what makes the difference here. 

Of course those parts are much faster than all before existing IGPs. But that can't be the criteria, right?

If you agree, I would move them back to integrated and disallow the usage in this stage.

 

[yosarianilives]

1 hour ago, Strunkenbold said:

I'm honestly not really happy with that move. Depending on who you ask, the definition of IGPs are sometimes GPU integrated in the CPU die or / and CPU and GPU die on the same package. So the Polaris GPU is on the same package with the kaby lake CPU. They are connected via a pcie bus, but it shouldn't be too surprising that they don't use something super fancy given the fact that a Intel part talks to an AMD part. 

Anyway I remember clarkdale having a GPU and CPU die. They were on the same package. Would anyone argue that it's not an IGP?  So I fail to see what makes the difference here. 

Of course those parts are much faster than all before existing IGPs. But that can't be the criteria, right?

If you agree, I would move them back to integrated and disallow the usage in this stage.

 

Interconnect on Clarksdale is dmi (or whatever the equivalent was then) , on motherboards its FSb or htt. Interconnect on hades canyon is pcie, if you soldered a mobile 1070 instead of using an mxm you wouldn't call that integrated etc

[mickulty]

17 hours ago, Strunkenbold said:

I'm honestly not really happy with that move. Depending on who you ask, the definition of IGPs are sometimes GPU integrated in the CPU die or / and CPU and GPU die on the same package. So the Polaris GPU is on the same package with the kaby lake CPU. They are connected via a pcie bus, but it shouldn't be too surprising that they don't use something super fancy given the fact that a Intel part talks to an AMD part. 

Anyway I remember clarkdale having a GPU and CPU die. They were on the same package. Would anyone argue that it's not an IGP?  So I fail to see what makes the difference here. 

Of course those parts are much faster than all before existing IGPs. But that can't be the criteria, right?

If you agree, I would move them back to integrated and disallow the usage in this stage.

Clarkdale's IGP is also the memory controller.  I think that all IGPs in the last couple of decades appear to be part of chips that have other purposes as well (CPU, northbridge, chipset or BMC), whereas all discrete GPUs are separate chips.

To futureproof against chiplets I would also suggest that any GPU that uses exclusively system memory is an IGP, whereas a GPU that normally doesn't use system memory is not an IGP.

[Leeghoofd]

I would also suggest this direction Gregor,  

Integrated GPUs use system memory only, The ones with dedicated memory  I would call them e.g. Integrated(memory)

My biggest issue when wanting to design any IGP related contest is the same questions always pop up; can we use Iris Pro or Hades based CPUs,? To keep it competitive the answer will be a solid no. Now it is also silly if we would always exclude these, time after time...So why not give these a seperate category

 

 

[yosarianilives]

In the case of iris pro I'd say we can call edram system memory as the cpu still has access to it and benefits it, whereas hades canyon the hbm stack is truly dedicated to the gpu. Also the iris pro gpu is on die with the cpu currently. A further conundrum there's plenty of motherboard igps of yester year where they had basically a dedicated Ddr3 chip on board right next to the chipset so they don't use system memory but they're definitely an igp. So I think the best way may be to loosely define an igp as graphics on a chip that has multiple purposes OR graphics that use system memory in the case of potential chiplet design apus. 

Edited November 27, 2019 by yosarianilives

[mickulty]

31 minutes ago, Leeghoofd said:

I would also suggest this direction Gregor,  

Integrated GPUs use system memory only, The ones with dedicated memory  I would call them e.g. Integrated(memory)

My biggest issue when wanting to design any IGP related contest is the same questions always pop up; can we use Iris Pro or Hades based CPUs,? To keep it competitive the answer will be a solid no. Now it is also silly if we would always exclude these, time after time...So why not give these a seperate category

This makes sense.  It solves the problem that Vega M while 'dedicated-like' is still inextricably linked to the CPU.

The HD 3200, 3300, 4200, 4250 and 4290 IGPs support "sideport" memory according to wikipedia, these would also fall into this category.  There are probably also some much much older examples.

[MetalRacer]

Got a sub not registering for stage 5 for some reason?

https://hwbot.org/submission/4304410_

[yosarianilives]

22 minutes ago, MetalRacer said:

Got a sub not registering for stage 5 for some reason?

https://hwbot.org/submission/4304410_

Gonna tack onto this, we currently have 3x lga 1151 scores which I know one of them will be pulled out because lga 1151 v1=v2 but needs to be done manually. However the z170 score coexisting with a z270 score makes me suspect the ranking is broken for that stage as they should both be the db as lga 1151 v1

[yee245]

I suspect it is messed up somehow, since my submission for that stage hasn't shown up for a week and a half (though MetalRacer's would replace mine, as it's higher on the same socket), even though we only have 5 submissions that are registering, and it should allow 6, and we don't have an LGA 1366 score counted either.

It's probably because keeph8n has a submission on a R3E of 266MHz, which is essentially out-scoring mine and MetalRacer's, causing neither to be counted, but also has that M9A score of 463MHz that is counted. From stuff that had happened in a previous team event, I think he (or a mod) would need to go and delete that submission to "open the door" for other team members' scores to be counted.

[yosarianilives]

Oh yeah I've seen that happen before. Still doesn't explain 3x 1151 scores

Edited December 18, 2019 by yosarianilives

[Leeghoofd]

if Team USA wants to remove some scores plz have an internal discussion which ones as some people still have backups, The Bot can not decide which scores have to go, At the moment the code is still looking for the highest score per hardware and if user 1A does multiple subs on different platforms it will block lower scores from other users. Therefore some planning and internal communication is mandatory before submitting...

Edited December 19, 2019 by Leeghoofd

[superpatodonaldo]

what about subs with tess off and VALID FM link?

[Leeghoofd]

will be removed, the idea behind a valid link is that the OCer doesn't play with any driver setting.... if you spot something out of the ordinary plz report

Edited December 19, 2019 by Leeghoofd

[yosarianilives]

13 hours ago, Leeghoofd said:

if Team USA wants to remove some scores plz have an internal discussion which ones as some people still have backups, The Bot can not decide which scores have to go, At the moment the code is still looking for the highest score per hardware and if user 1A does multiple subs on different platforms it will block lower scores from other users. Therefore some planning and internal communication is mandatory before submitting...

It should only compare scores that are counting towards the comp no? So if there is currently no 1366 score counting towards the comp, then a 1366 score that doesn't count should block any scores.

Also doesn't answer what's going on with 1151 as the z270 score should block a lower z170 score but it doesn't currently for our ranking. So we have 3x lga 1151 scores counting towards our total. I think the ranking may have entirely broken for this one.

EDIT: Found what the problem is with the ranking. The "z170" score that's counting is actually a z97 score because the maximus VII extreme was mis-subbed as Maximus VIII extreme. However this exposes a flaw to the rankings, rankings are supposed to be done by Motherboard socket however they are apparently being done by CPU socket. This also means that LGA 771 can be done on a lga 775 board just fine according to the way the ranking is configured, despite the ranking supposedly being explicitly set up to prevent this.

Edited December 19, 2019 by yosarianilives

[darco_2]

Hey guys. 

Got a problem with submiting for CC. Can't submit, in any way. Although my score got registered and ranked on hwbot, when i click the participate button i get the "sorry, an error occured. Are you shure you meet the requirements?" error. All the requirements are met in my case. This problem startet 10-12 days ago and, since then, can't submit any score. Any ideas what is happening? 

[keeph8n]

submit from within the competition page

[Leeghoofd]

@yosarianilivesIf the software also has to correct the end users inability to do a proper submit we would have to wait for a higher level of AI to get it working.

Just stop being such a loophole whore dude.. It is not because the software can't be 100% configured that you must abuse it.... 

 

For the 3Dmark scores with a valid link and a suspicion of Tesselation settings being altered, these will be properly moderated over the weekend....

[yosarianilives]

1 hour ago, Leeghoofd said:

@yosarianilivesIf the software also has to correct the end users inability to do a proper submit we would have to wait for a higher level of AI to get it working.

Just stop being such a loophole whore dude.. It is not because the software can't be 100% configured that you must 

This has nothing to do with loopholes, the stage is misconfigured. It is dividing scores by cpu socket when it's supposed to be configured to divide scores by motherboard socket. As for the issue of 1366 where a score that doesn't count for the stage blocks another score, we solved it by manually deleting scores that don't count towards the total. 

[darco_2]

7 hours ago, keeph8n said:

submit from within the competition page

I did try that also. Get the page "ooops, an error ocured" 

[Leeghoofd]

Like stated before the Bot cannot detect which score to use if one user subs multiple subs outscoring the other ones of his teammates on sort like platforms. 

Dividing by mobo sockets has been broken for ages. So this stage requires some special attention by the mods, why did you think I added 1151 is the same as 1151v2? 

Anyway all subs in this stage look good so far, so no problem. 

[superpatodonaldo]

7 hours ago, Leeghoofd said:

For the 3Dmark scores with a valid link and a suspicion of Tesselation settings being altered, these will be properly moderated over the weekend....

as far as I remember that score was already removed: https://hwbot.org/submission/4300610_jumper118_3dmark___fire_strike_ultra_radeon_pro_duo_(fiji)_9305_marks/

also this run seems to have tess off: https://hwbot.org/submission/4286480_triox.sgi_3dmark___fire_strike_ultra_radeon_pro_duo_(fiji)_9452_marks

Edited December 20, 2019 by superpatodonaldo