One of the Biggest Cable Companies Says Cable TV Isn’t Working

One of the Biggest Cable Companies Says Cable TV Isn’t Working

It’s our newest large Mod Mat at 4 feet by 2 feet or 122 by 61 centimeters
for PC building and parts modding.
We just received this at the warehouse alongside our GN15 shirts,
and both will be shipping out soon.
The limited run Mod Mat has the GN15 logo centrally
plus a host of new diagrams to make PC building and maintenance easier.
New helpful images include the most common motherboard front panel wiring layouts
we’ve seen for LEDs and buttons,
8P8CRJ45 wiring diagrams for those of you building your own Ethernet cables,
some helpful angular and length references,
and more.
We also just included stuff that’s cool,
like a Zen 4 CPU die location diagram for its unique die layout and locations,
the simplified compositional cross-section of the same type of CPU,
and some quick references for typical fan orientation to help new PC builders.
The screw sorters on the left feature two GPU silhouettes for screw tracking
for your water block or re-pasting projects,
plus a row of alternating color blocks to help keep mental track
of where different parts came from.
The DDR modules are also to scale in the top right.
These are also high heat resistance mats,
making them great for tube bending and working with heat guns.
If you liked our recent CPU sample size testing
and you find today’s video helpful for your build,
please consider buying one of the large mod mats
to help support our work and get something useful for your projects.
These GN15 mats are rugged,
durable surfaces
that can take a beating and protect your table
and your components from dings and scratches.
Grab one on store.gamersnexus
.net
today
and help us celebrate 15 years of running benchmarks like this.
Okay,
so some context.
Whenever we’re benchmarking a brand new game that rolls out like this,
especially when it comes in as hot as Starfield did,
we immediately set forth trying to collect some data on
how does it just perform in general and different environments of the game.
So we’ll be talking about that a lot today.
Came in early access at 8 p.m.
Patrick and Jeremy set up the benches for me during the day.
I shifted my schedule,
basically came in right when it launched,
worked through the night,
and then Jeremy and Mike came in early
to help me finish up the testing and the editing.
So to my team who enabled all of this
and my overnight obsession benchmarking,
thank you.
We’ve got some cool content.
All right.
So these tests used the latest NVIDIA and AMD Day One drivers.
We have fresh bench-to-table benchmarks.
And again,
Intel’s not on here.
It doesn’t work yet.
There are a lot of different ways to test games,
and so there’s going to be a ton of room for differing results.
As for our approach here,
the choices we made included disabling FSR,
disabling VSync,
and also disabling variable rate shading as well as dynamic resolution.
These will impact results compared to tests that might leave them on.
So you should look at the numbers largely against themselves
and avoid cross-comparing if there are unknown variables in the settings.
Our reasoning for turning those things off
was mostly just to control the test environment.
So FSR,
for the most part,
it’s fairly predictable scaling.
So we’re testing at native,
and then from there,
what we care about is the relative distance one card to the next.
We might revisit this with additional specific FSR testing in the future,
but since things are kind of in flux right now with the game and drivers and everything,
we kept it very simple with just native resolution,
and then we can extrapolate findings from there.
As for the rest of it,
standard to turn VSync off,
and variable rate shading does introduce variables as well.
So for the rest,
we stuck with fairly simple high and ultra settings.
We have some medium and low in there as well,
and we ensured that dynamic resolution features were all completely off,
tested at 1080,
1440,
and 4K in the old-school method,
which is just the resolutions as they are.
Our goals for finding a test location primarily revolve around a couple of things.
One is repeatability,
of course.
So we can’t have too many random elements like surprise combat,
for example,
happening in the test scene because we need that reliability.
The second point is that the test location should be close to a worst case,
but ideally,
it’s generally representative of gameplay.
So it’s not the absolute worst case possible.
It’s more of like a heavy load,
like a city environment,
for example.
That’s what we end up choosing.
It must still be a common gameplay scenario,
in other words,
but still a heavy workload.
For this process,
we captured the frame time,
so the first couple of hours of play,
and logged them.
This included the tutorial cave,
the initial outdoors location on the moon,
or asteroid,
or whatever it was,
the ship,
takeoff,
space flight,
combat on foot,
and combat in space,
Crete and its indoors research lab,
and New Atlanta City.
Games are highly variable,
dependent on location,
and choosing a location has all the impact on the results
that sets us up for long-term benchmarking.
This research was captured on a 4060 Ti at 1440p and high.
We want to emphasize here that the settings on the card are basically irrelevant
for the research stage,
other than one important thing.
Prior to getting all of the GPUs involved,
we start on one card,
and that card’s job is to be GPU-bound.
We want it to be around 50 or 60 or so FPS,
generally speaking,
assuming that there’s sufficient CPU headroom,
and there is here,
and that’s so that we can find locations that show scaling.
So,
ideally,
we want to be able to see a wide range,
assuming the game has a wide range of types of play,
like non-combat,
combat,
Dense citiies,
and open areas,
and space,
in this case.
So,
for this,
we choose settings that typically put us closer
to that sort of middle 50,
60 FPS range,
and then we can scale from there.
Here’s the chart.
We have FSR upscaling,
dynamic resolution,
and variable rate shading all disabled here,
so there’s no FSR upscaling at all.
The 4060 Ti at 1440p high held in the 70s in a few locations,
notably inside one of the areas of the early cave and inside the lab on Crete.
And again,
the importance here isn’t the absolute frame rate of the 4060 Ti,
because if you don’t have that card or want it,
then that’s not a helpful number.
But what is helpful is the range.
So,
we observed in the 40s FPS average for initial conversations,
where high poly detail is loaded in for characters’ faces.
Outdoors roam in ranged,
but we saw mostly in the 40s there as well.
Space combat and space flight had us typically around 50 FPS,
plus or minus a bit.
Takeoff had a severe frame time spike a couple of times,
actually,
but as this moves to a hands-off animation,
that just doesn’t seem as important as,
say,
in actual gameplay.
We also captured two sets of 10 minutes of gameplay shown here in New Atlanta City.
These numbers are useful to give us a fuller picture of averaged frames across large parts of the game,
with loading screens removed from the numbers,
because those are pointless.
These should also be considered fairly representative of play.
We’re going to spare you all the details of going over each of these numbers individually and talking about it,
and if you want to learn more about sort of our selection process for benchmarks,
you can check out our sample size test for CPUs recently.
We show a lot of the behind-the-scenes for that.
But there are two important entries at the bottom.
One is called All Gameplay Average.
That’s the average FPS,
and 1% of each number above it.
In other words,
we’re producing a common frame rate of 53 across all scenarios tested,
including averaging for the tests in the 70s.
And that’s across the first few hours of play by measuring in different areas.
Once we had this number,
we chose an area of New Atlanta City that was fairly close,
but leaned slightly heavy.
That way,
we have a representative sample of a higher-load city,
and the end result was 48.6
FPS average here.
Remember that it’s possible there are areas later in the game that are heavier.
But given that we captured,
again,
a couple hours of play and came up with these numbers,
we think the test area selected is a safe,
sort of lower-end frame rate representation
for high-density areas without being ridiculous.
We tested in the residential district of the city and had a fixed walking path.
We did a number of one-off tests in this area.
But before we get into the comparative charts,
let’s look at some quick settings research.
As the next part of our research process,
we started using our actual bench pass.
And for this,
we tested each preset with the one research card,
with,
again,
FSR disabled,
all types of dynamic resolution turned off,
and VSR or VRS turned off as well.
And this was in the same location,
using the same test,
and we have a minimum of four passes here.
We didn’t need to discard very many.
We did run additional for some of these cards,
though—upwards of six or seven for a few of them.
Generally speaking,
we were within about one FPS standard deviation run-to-run.
Lows are always wider here since there’s less data to work with by their nature.
But the averages are extremely consistent from one run to the next,
making this overall relatively easy to parse
and meaning that we didn’t need to discard or,
as stated,
rerun many tests.
So that’s good.
Up first,
our 1080p high chart.
This is the densest one.
And remember,
we did all of these cards in,
like,
12 hours or so.
So this will get us started pretty well,
and then we’ll keep adding to these as we approach the 7800 and 7700 XT reviews.
In this one,
the RTX 4090,
the RX 7800 XTX,
and the RTX 4080 can all be ignored.
They’re all bottlenecked.
These just didn’t have enough scaling room.
You’re not seeing a difference between them because there’s not any room to see one.
We’ll have to wait till we get to 4K to start seeing that,
and we’d need either a much higher-end CPU with a higher clock
or an area of the game with less CPU load.
This has some crowd density to it.
Shifting attention to unbound comparisons,
the 6800 XT remains a strong modern option,
and it sits atop the 4070 here,
with the 6700 XT technically leading the 4060 Ti,
but they are functionally equivalent.
The RTX 3080 remains completely viable here as well.
Not much of a surprise.
It’s not really…
not old,
and it was a good card.
The 3060 Ti and the 4060 Ti have more of a gap here than they have at higher resolutions,
but that’s behavior we already saw in our 4060 Ti review.
They tend to converge,
and the last-gen option can even surpass the 4060 Ti in some games at 4K resolution.
The RX 7600 and the RTX 4060 are roughly tied in average frame rate,
with the 7600 having a technical lead and a price advantage,
at least right now.
As for older hardware,
like the 1070 and the 2060,
we’d have to turn to lower settings for these.
We added these at the last minute just because we know a lot of people are still on them,
and it’s nice to see the older hardware in modern games.
Now,
we didn’t test many cards at 1080p medium,
but here are the ones we did.
The movement and absolute FPS isn’t much.
The 1070 and the 2060,
of course,
benefit from the drop in settings,
but not enough to move the needle without other settings reductions as well.
The RX 7600 overall does well relative to the rest of this group,
especially with its newer sub-$250 price that we’ve been seeing around.
We only ran two cards at 1080p low.
Here are the results for them.
The 4060 gained about 10 FPS against its test at medium settings previously,
with the 2060 gaining a little less than that.
Moving to 1440p and high settings,
the top remains bottlenecked,
although it’s reducing a little bit.
The equivalent FPS between two resolutions for the top three cards
indicates that bottleneck pretty clearly to us.
The more relevant comparisons happen below the 95-plus range.
The 4070 and the 3080 are about equal here,
with the 4060 Ti and the 6700 XT also close in rank.
None of the lows in this chart jump out to us as particularly bad.
In general,
they’re not keeping pace with the average as well as we’d like.
Normally,
that red and blue bar would be a little closer together,
and we’ve seen that in other tests in the past,
but they’re consistently sort of disparate across all devices.
That gap remains true.
It’s not like it’s any one particular card that is the result of it.
It’s just the game at this point.
We’re on to 4K high next.
We didn’t do as many tests on this one as the prior settings
due to the intensity of the configuration.
The RTX 4090 predictably leads the chart,
but it’s biotechnicality.
By reality,
it’s tied with the 7900 XTX.
They’re within run-to-run variance of each other for average 1% and 0.1
% lows,
which is actually really impressively close.
The RX 7900 XT manages 87% of the performance of the 7900 XTX at 64 FPS average.
The RTX 4080 is just behind that at 58 FPS average,
which means that the gap between the 4080 and the 4090 is 30% here.
That’s about the same as we’ve seen in other games dating back to the 4080 review.
The 3080 is also near its usual distance from the 4090,
although a little higher than we’ve typically seen.
It allows the 4090 a little over an 80% lead.
It’s not unheard of,
but it’s a little on the high side.
You can check our old 4090 review for more of that.
As for all these cards under the RTX 4080,
remember that we’re one,

and two,
establishing scaling in a relative sense.
It matters more than the absolute FPS for the way we run benchmarks.
You can reduce settings or add upscaling
and increase the frame rate to good playability
on something like a 6800 XT here or a 3080,
and with FSR,
even lower than that.
But for testing purposes today,
we’re showing the ranking of the cards relative to each other.
The 1080 Ti once was among the few early cards
that could really handle 4K games well,
but for this one,
it’s just not able to do that,
at least not without external assistance from modern upscaling techniques.
And to be clear,
it would probably be kind of rough even then.
We haven’t tested that on the 1080 Ti yet.
The game’s still too new.
4K Ultra is pretty simple.
We don’t have many cards that can run this well,
so it’s a limited list.
The RTX 4090 ran 4K Ultra without upscaling
and in the city at 68 FPS average,
with lows paced roughly proportionally behind.
That has the 7900 XTX and the 4090 once again about tied,
with no meaningful difference between them
in any of the three reported metrics.
The 7900 XT allows the 7900 XTX a lead of 14%,
while the XT non-second X leads the 4080 by about 14%.
Again,
the 4080 leads the 6800 XT here by 26%.
Now,
even at the 6800 XT level,
you could still run at 4K Ultra
with some use of,
once again,
FSR and upscaling.
It’s an option here.
It’s not a particularly great experience at 40 FPS,
but it’s not far off from something that would be fine.
And dropping settings or adding FSR
would get it the rest of the way there,
if that’s really important to you.
So,
recapping some interesting findings.
First of all,
the reminder,
once again,
repeated it three times,
sorry,
but this is important.
This is day one,
and so we push really hard
to get as much good data out as we can early
to help people make those early decisions.
It is going to change.
That means if you look at this video
in six months from posting,
you should probably check our channel for a follow-up
or for the most recent GPU or CPU review for that period
because this situation with drivers and game software,
it’s going to change.
But the data is still really interesting,
and the data itself may remain relatively scaled
the same as it is today.
We’ll see.
Right now,
AMD has very competitive performance
in some cases.
So,
as compared to the 4090,
it’s actually particularly competitive.
As compared to the 7600 and the 4060,
they have a strong foothold there
compared to some of the other games we’ve tested in the past.
And that’s interesting.
It may be the case that this is a jump that AMD had
with thanks to their exclusive partnership.
It could also be the case that this particular game engine
or the way it’s built in general
is benefited by the hardware,
the architecture for AMD’s RDNA 3,
more so or disproportionately so versus NVIDIA.
And we see that sometimes where one game versus another,
you might have a 15% swing or something
in the numbers based on how it behaves
on that particular architecture.
It could be a memory thing.
It could be bandwidth.
It could be ROPs.
There’s lots of options there.
Intel will revisit as soon as they have a driver.
So,
we’ll have news for you soon on that,
probably about a week.
And as for the game itself,
it’s pretty taxing.
So,
just strictly speaking performance,
not talking about optimization
or subjectively what we think of the graphics or anything.
Strictly on the hardware,
it is a taxing game.
The top cut of cards are bottlenecked on other components

say,
1080p and 1440p,
yes.
But older cards like a 1080 Ti,
2060,
they struggle a little bit more than we would expect at 1080p.
And you get into a situation where,
at least right now,
with the current game version as of Friday the 1st

you know,
middle of the day Eastern,
which becomes relevant because who knows when they’ll patch it.
Right now,
those cards need some help.
It could be upscaling.
It could be a reduction of graphics.
But they need some help to really run this well.
Now,
there’s no FSR 3,
unfortunately.
There’s FSR 2 in there.
There’s no DLSS yet.
I think everyone knows that at this point.
And the Intel Arc drivers are kind of an interesting trail to follow as well
because we’re not sure if that’s an Intel,
like the hundred plus billion dollar corporation thing,
or if that’s a Bethesda thing,
as in Bethesda slash AMD.
Nvidia has it working on their cards.
It doesn’t appear that they have optimizations or many of them yet.
We did use the game ready driver.
Quick cursory glance,
it looks like maybe the same as the previous driver.
We haven’t 100% confirmed that.
It’s been a long night of testing.
But the fact that Intel doesn’t have day one support is disappointing.
We just don’t know which party to be disappointed at right now.
It’s also not totally unexpected.
You kind of know what you’re signing up for with Arc.
Everyone’s talked about it at this point.
All right,
a couple things.
So the settings,
we wish they had an FOV slider.
It was one of the things I noticed right away.
It had kind of limited controls.
You don’t have many graphics options.
And in terms of expanding that,
it’s probably going to come from mod support.
The findings we had throughout the game were pretty interesting.
So in that sort of case study with the 4060 Ti,
seeing the 20 FPS range in conversational dialogue
versus,
say,
indoors combat in lower density areas,
that was pretty useful for understanding
the plus or minus versus the benchmark area we chose.
And we’ll keep looking around for more variable performance in the game as well.
For now,
though,
that’s it.
I’m going to go to bed and get this video live.
Not in that order.
And we’ll be back shortly with more Starfield content.
Really interesting game to test.
It’s an exciting time right now.
There’s a lot of cool games coming out or that have come out,
and we’re excited to run the numbers on them.
So check back for more of that.
Subscribe for more.
Go to store.gamersaccess
.net
to grab our brand new GN15 Mod Mat.
It is for the 15-year anniversary.
We have a lot of cool diagrams and graphics on it.
It’s very helpful,
screw tracking,
GPU silhouettes for disassembly,
water block and stuff like that.
So grab one if you want to support us.
Thanks for watching.
We’ll see you all next time.

 

Travis Barker Leaves Blink-182 Tour for ‘Urgent Family

Leave a Comment