CFM = Cubic Inches x RPM x Volumetric Efficiency ÷ 3456. Any ordinary stock engine will have a volumetric efficiency of about 80%. Most rebuilt street engines with average bolt-ons have a volumetric efficiency of about 85%, while race engines can range from 95% up to 110%. Example: Using a 355 CID engine x 5,500 max rpm = 1,952,500 Take 1,952,500 x .85 = 1,659,625 Then 1,659,625 ÷ 3456 = 480 CFM Even with about a 10% cushion, a 500 CFM carburetor will handle this engine great. https://www.speedwaymotors.com/the-toolbox/what-size-carburetor-do-i-need/28634 I built my first SBC in 1980, and back when I was sixteen, the bigger the carb the better... (We thought) Over the years, the struggle to get those big carbs to run right, eventually taught me that it was easier to get a smaller carb to run "right" than it was bigger carb. Jetting the bigger carbs down is harder than jetting the smaller carbs up. It is usually not the airflow that is the issue, it is the fuel to air ration, along with the accelerator pump (squirter) volume that needs tweaked. The last 434 I built had peak power at just over 7000 rpm, so works out to need 747.1938 cfm carb. Using this formula and then tweaking a 750cfm Black Diamond very quickly brought the AFR into line where we wanted it. https://www.holley.com/products/fue...l/black_diamond_series/q-series/parts/BDQ-750 I have half a dozen old carbs on the shelf from those days when bigger was better. I have literally not touched, or found a good use for ANY of them in close to 20 years now. Don't be afraid of your 670. I am thinking even it might be too much carb, but it should be possible to make it work splendidly with a little tinkering. Use the closest thing you have, find someone with an AFR gauge, or buy one and install it. I have one on my break in stand, and have found it to arguably be the most important thing I ever bolted on that stand. https://www.amazon.com/AEM-30-4110-UEGO-Ratio-Gauge/dp/B00N3VGPYS You will be WAY happier with the results, and so will your engine! No more guessing.