Intel Core 9x00 Processor Prices Listed at Singapore Distributor

#5584024 Posted on: 09/13/2018 07:49 AM

---

A video card that offers 30-45 frames at 1080P while using ray tracing, all for 1200 bucks US ? I don't think anyone wants that.

#5584030 Posted on: 09/13/2018 08:37 AM

---

Completely agree.
Yet we see 27" monitors cost £2200 and mainstream GPU £900-£1200, and still people buy them instead of boycott them.
And anyone who complains against it, is a hater and competitor fanboy. Yet not even 1/5th of them actually buy the product.

Pure religious fanatics.



That 2700X was expensive. in UK is sold (Amazon) for £285 (2700X) and £252 (2700) which includes 20% VAT (sales tax).

And I agree with you about fine tuning. Bought a new Vega 64 Nitro+ in July (2 8-pin), spend 20 minutes tinkering the settings in wattman (downvolting/overclocking), and while is consumes 280W, it beats by country miles the performance of the GTX1080 @ 2190 i had previously. And I am not the only one who used those settings, others achieved the same perf on the Red Devil also.

Also that shows, how half a*sed the reviewers are when comes to Vega GPU reviews.
You do not review and compare GPUs overclocked. That's not fair towards masses. HH for example shows for each card OC potential separately. Apparently, without editing vBIOS and bumping power limits. But showing OC potential is not about showing maximum performance.

#5584145 Posted on: 09/13/2018 04:08 PM

---

I'm pretty sure AMD will kick Intel in the teeth when these puppies release. AMD are gonna rub salt in that wound and lower their 2600 and 2700 prices even more.

I have it on good authority that this will 100% happen and etailers have already been prepped on this.

Spot On!
and again, due to marketing being at the helm at Intel, not only is 10nm the most expensive vaporware in history, but they can't even supply 14nm (to the appropriate market saturation of a company as big as Intel) as they have to farm out the chipsets to their biggest competitor in the fab business (TSMC)...who by the way, has stolen the crown of world's largest chip fab.

etailers and retailers have no choice but to stock AMD as they will actually have products in the warehouse ready to sell.

and i began divesting my Intel portfolio as soon as i heard the TSMC news.

if Lisa Su worked for Intel, AMD would be dead already.

#5584151 Posted on: 09/13/2018 04:22 PM

---

Haha yeah lets trust AMDs own numbers. Sigh.

Zen 2 comes out in Q2-Q3 2019. Intel will release 10nm in Q3-Q4.
Who I do not trust is some random guy who knows AMD's and intel's release dates better than them.
Do you think someone believes you when you push intel's King this/that? Intel's IPC improvements over years became pretty poor, and fact that they are releasing double amount of cores for desktop than they did before Ryzen tells story of how much they could improve IPC in that time.
If lucky, intel mitigates impact of Spectre/Meltdown fixes.

On other hand, Zen1 is completely new architecture. Anyone who thinks that there is not enough of low hanging fruits to deliver at least 8% IPC improvement in more than 1,5 years since original release of architecture does not know anything.
And same goes for 7nm effect on clocks.
Ryzen 2700X with all cores on 4,25GHz eats 180~200W under full load depending on type of workload. that's 75~95W above rated 105W. Mere power efficiency from 7nm will move such chip from ~1750 to 1950 points in CB-R15 in 105W limited scenario. 5% higher clock which is expected will move max from 4,3 to ~4,5GHz. It is that simple. Those are very low exectations (baseline, if you want).

As far as clock improvements go. From my experience, increasing multiplier did not yield stable OC above 4,35GHz at all. But increasing BCLK allowed me to go to 4,48GHz (w/o Voltage bump). This means cores themselves did not have problem, it was problem with mechanism controlling clock. If AMD fixes that... Looks like damn low hanging fruit to me.

#5584170 Posted on: 09/13/2018 05:20 PM

---

Oh really. Glad you're mad. Now accept facts:



Add to this that HT allows for higher clocks because of lower temps...

If you want more proof, Visit Gamers Nexus. They include HT OFF in their tests and it beats HT ON 99 out of 100 times clock for clock.

You learned something today. Creating that account was worth your while.

caught my attention to this.... well i not type that OC things or tweak to maxed out the performance
so never care much about few-fps different

saw the vids, nothing worth to notice the difference... probably only like 3-5fps difference? but that also depends on the game, BF1 is better with HT on than off

so in-term games perform better with HyperThread OFF, its probably right
but dont think most people care about those minisucle different

and if there big difference with HT on and off, i believe its the game issue rather than HT causing the issue

#5584178 Posted on: 09/13/2018 05:45 PM

---

People turn off HT/SMT to get higher clocks no?

I thought the whole reasoning behind turning HT off was to achieve better clock speeds on the actual cores.

I'm not sure about Zen but wondered if you have 1800x 2600x/2700x and turned off SMT can you get higher clocks on those actual 6/8c procs?

I know you might be able to save energy that way but i'm sure hardly anyone would do that unless it's proved to increase clocks hence FPS.

#5584192 Posted on: 09/13/2018 06:45 PM

---

Vega 64 Red Devil and Nitro+ have factory set clock to 1630. However without downvolting this speed is never achieved. The card depending thermal will work at around 1480-1540.
By spending 3 minutes reducing the GPU P6 & P7 voltage by 170mv (-12%) without touching the clocks, the performance jumps quite a lot alongside the clocks. (no overclocking). If you set P7 from 1630 to 1655, the card jumps speed to 1620s (not even factory clocks) while operating almost similarly to a GTX1080 @ 2164 - 2190 overclock.
And all these at 270-280W. Nowhere near the 330-370W Guru3d has on their review.....

#5584212 Posted on: 09/13/2018 07:40 PM

---

Vega 64 Red Devil and Nitro+ have factory set clock to 1630. However without downvolting this speed is never achieved. The card depending thermal will work at around 1480-1540.
By spending 3 minutes reducing the GPU P6 & P7 voltage by 170mv (-12%) without touching the clocks, the performance jumps quite a lot alongside the clocks. (no overclocking). If you set P7 from 1630 to 1655, the card jumps speed to 1620s (not even factory clocks) while operating almost similarly to a GTX1080 @ 2164 - 2190 overclock.
And all these at 270-280W. Nowhere near the 330-370W Guru3d has on their review.....
I know it since I used to have Fury X. VBIOS: Undervolting and for fringe scenarios bumping all power limits. W/o any OC I had higher performance than many reviewers with OC.
Most tragic reviewers had actually same or lower performance than they had at stock clock.

But it still does not change fact, that it is AMD's responsibility to set those voltages. And good reviewers will deliver result for power states which you get upon buying that cards.

#5584388 Posted on: 09/14/2018 09:53 AM

---

this seems very interesting if only intel didnt take the foot of the pedal

#5584654 Posted on: 09/14/2018 09:11 PM

---

Watch the vid again, HT off is clearly better. Also, this test is done at same clocks. HT off generally allows for higher clocks because of lower temps (more headroom). HT OFF + 200 MHz on all cores and we're talking +10% overall.

Ryzen with SMT off performs better in games too. This is nothing new.

HT/SMT only helps when there's too few physical cores. Helps alot on i3 for example. Also older i7 quads, in a few games. In most games 4 cores is still enough. 6 cores is enough for everything.

i7-9700K is going to be an awesome gaming chip. High fps gamers will have no reason at all to even consider the i9-9900K, unless they plan to use it for the next 10 years.
So, do you run your CPU with HT OFF?

#5584915 Posted on: 09/15/2018 04:43 PM

---

Thanks for the laugh guys. Almost made me spill my coffee.